{ "schema_version": 1, "published_date": "2026-10-07", "hardware": { "gpu": "EVGA GeForce RTX 3090 OC FTW3 Ultra", "board_variant_source": "Owner supplied; NVIDIA model/VRAM independently read on 2026-10-07", "nvidia_reported_name": "NVIDIA GeForce RTX 3090", "vram_mib": 24576, "driver": "595.91.07", "cpu": "AMD Ryzen 9 8945HX", "physical_cores": 16, "logical_threads": 32, "ram_nominal_gb": 96, "linux_MemTotal_kib": 94397108, "linux_MemTotal_gib": 90.02409744262695, "nvme": "KINGSTON SKC3000D4096G", "nvme_nominal_tb": 4, "kernel": "7.0.0-34-generic", "os": "Ubuntu 26.04.1 LTS", "folder_note": "GPU corrected by owner: RTX 3090, not Ti. Directory name corrected accordingly." }, "historical_snapshot": { "source": { "path": "jarvis-flash174/report/benchmark-catalogue180.json", "bytes": 2948385, "sha256": "e5b053ee7bbaab59baae2fb57f8fc4defd1990e1056d2937a9071df7861b16d7" }, "totals": { "runs": 82, "captured_requests": 339, "native_benchmark_runs": 32, "native_result_rows": 56, "native_recorded_samples": 168, "profiling_runs": 3, "captured_configurations": 49, "raw_files": 2217 } }, "runs": [ { "id": "context176-262144-cpu48-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T13:55:19.758563+00:00", "ended_utc": "2026-10-05T13:56:57.890873+00:00", "wall_seconds": 98.13233597297221, "case": { "name": "ctx262144-cpu48-q8_0-b8192", "context": 262144, "cpu_moe": 48, "kv": "q8_0", "cache": 0, "batch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "failure": { "category": "QAError", "message": "QAError: CUDA/assert error in own child log" }, "resource_guard_failure": "QAError: CUDA/assert error in own child log", "depths": [ 65536, 131072, 196608, 235929 ], "speed_repeats": 3, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C001" ], "source": { "path": "context176-262144-cpu48-q8-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, "summary_source": { "path": "context176-262144-cpu48-q8-20261005/summary.json", "bytes": 5731, "sha256": "4ebf123dc08261f208ca6ec14c7daf11855d039e5eb086ecf8430370fb6ca666" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 78, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 4228.0, "board_vram_min_free_mib": 19899.0, "host_ram_min_available_kib": 39689728.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-262144-cpu48-q8-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T14:04:29.948164+00:00", "ended_utc": "2026-10-05T14:05:51.149914+00:00", "wall_seconds": 81.20177861896809, "case": { "name": "ctx262144-cpu48-q8_0-b8192-u1024", "context": 262144, "cpu_moe": 48, "kv": "q8_0", "cache": 0, "batch": 8192, "ubatch": 1024, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "failure": { "category": "QAError", "message": "QAError: Finite QA signal SIGTERM" }, "resource_guard_failure": "QAError: Finite QA signal SIGTERM", "depths": [ 65536, 131072, 196608, 235929 ], "speed_repeats": 3, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C002" ], "source": { "path": "context176-262144-cpu48-q8-u1024-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, "summary_source": { "path": "context176-262144-cpu48-q8-u1024-20261005/summary.json", "bytes": 5737, "sha256": "76641ff146d06b18215e41dcad101da9b6ef6fb0ba143165a17e6b26e59634e7" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 60, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 4136.0, "board_vram_min_free_mib": 19991.0, "host_ram_min_available_kib": 85692960.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T14:09:25.606885+00:00", "ended_utc": "2026-10-05T15:01:03.300714+00:00", "wall_seconds": 3097.6938545739977, "case": { "name": "ctx262144-cpu48-q8_0-b8192-u2048", "context": 262144, "cpu_moe": 48, "kv": "q8_0", "cache": 0, "batch": 8192, "ubatch": 2048, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 65536, 131072, 196608, 235929 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C003" ], "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "summary_source": { "path": "context176-262144-cpu48-q8-u2048-20261005/summary.json", "bytes": 7186, "sha256": "7861713386491472033365b7ed17d46b12127c87fe29bd94cc32257aa9ad2173" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 3087, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21732.0, "board_vram_min_free_mib": 2395.0, "host_ram_min_available_kib": 26370840.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T13:40:12.539831+00:00", "ended_utc": "2026-10-05T13:50:05.469811+00:00", "wall_seconds": 592.9300139979459, "case": { "name": "ctx32768-cpu43-q8_0-b8192", "context": 32768, "cpu_moe": 43, "kv": "q8_0", "cache": 0, "batch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 8192, 16384, 24576, 29491 ], "speed_repeats": 3, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C004" ], "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "summary_source": { "path": "context176-32768-cpu43-q8-20261005/summary.json", "bytes": 6866, "sha256": "7fcbe1ff12affeb72174e9b745ea11909e14b4860c88e6b5d2bc339ab793d7c2" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 584, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21088.0, "board_vram_min_free_mib": 3039.0, "host_ram_min_available_kib": 37891440.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-32768-cpu48-q8-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T13:52:16.268604+00:00", "ended_utc": "2026-10-05T13:55:19.702468+00:00", "wall_seconds": 183.43389157298952, "case": { "name": "ctx32768-cpu48-q8_0-b8192", "context": 32768, "cpu_moe": 48, "kv": "q8_0", "cache": 0, "batch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 8192 ], "speed_repeats": 3, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C005" ], "source": { "path": "context176-32768-cpu48-q8-control-20261005/requests.json", "bytes": 34850, "sha256": "ff498936a7091a684a8353a86fd943613ec9e70fa85f1e8ba5db3637f7977a56" }, "summary_source": { "path": "context176-32768-cpu48-q8-control-20261005/summary.json", "bytes": 6036, "sha256": "2aebac4f7f51cb2bcd4f6798beb53b8d93617f8bb7e1c0cb7c65fb39acd48546" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 172, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16286.0, "board_vram_min_free_mib": 7841.0, "host_ram_min_available_kib": 33733232.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-32768-cpu48-q8-u1024-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T14:00:46.030033+00:00", "ended_utc": "2026-10-05T14:04:29.894088+00:00", "wall_seconds": 223.86409041297156, "case": { "name": "ctx32768-cpu48-q8_0-b8192-u1024", "context": 32768, "cpu_moe": 48, "kv": "q8_0", "cache": 0, "batch": 8192, "ubatch": 1024, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 8192 ], "speed_repeats": 3, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C006" ], "source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.json", "bytes": 42460, "sha256": "d0f8d9927daefdc84aee21bf558e7c953254a9a823ba1fc860270028e4019d02" }, "summary_source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/summary.json", "bytes": 6069, "sha256": "efca0c31366fff671e7ad9c9bec7370227d3710ee6008fcc87106d7630ccfddb" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 213, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 9604.0, "board_vram_min_free_mib": 14523.0, "host_ram_min_available_kib": 36322408.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-32768-cpu48-q8-u2048-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T14:07:05.339230+00:00", "ended_utc": "2026-10-05T14:09:25.546217+00:00", "wall_seconds": 140.20701211097185, "case": { "name": "ctx32768-cpu48-q8_0-b8192-u2048", "context": 32768, "cpu_moe": 48, "kv": "q8_0", "cache": 0, "batch": 8192, "ubatch": 2048, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 8192 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C007" ], "source": { "path": "context176-32768-cpu48-q8-u2048-control-20261005/requests.json", "bytes": 18057, "sha256": "466962bb262f626bd1930f939d96a9dd005892e3a93d77f30ff95dec62b558c0" }, "summary_source": { "path": "context176-32768-cpu48-q8-u2048-control-20261005/summary.json", "bytes": 6069, "sha256": "4a745010e05f6fffe80575588f4c49c6c5275af0d019485a4e64c89f3c5fe477" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 129, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10376.0, "board_vram_min_free_mib": 13751.0, "host_ram_min_available_kib": 36022080.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-40960-cpu44-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T15:12:34.166615+00:00", "ended_utc": "2026-10-05T15:16:11.615618+00:00", "wall_seconds": 217.449030985008, "case": { "name": "ctx40960-cpu44-f16-b8192-u8192", "context": 40960, "cpu_moe": 44, "kv": "f16", "cache": 0, "batch": 8192, "ubatch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 36864 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C008" ], "source": { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/requests.json", "bytes": 38009, "sha256": "19a64e38710db82a54606b61e0ab7264ed9d57f6e11337b5bb20873e2b9edf41" }, "summary_source": { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/summary.json", "bytes": 6334, "sha256": "91386b2a8101d0f7891f838b03169a1a811950857f55baf4d54a0848a4fb42e8" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 208, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21672.0, "board_vram_min_free_mib": 2455.0, "host_ram_min_available_kib": 35980600.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-45056-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T15:16:11.673022+00:00", "ended_utc": "2026-10-05T15:19:56.499991+00:00", "wall_seconds": 224.82699551898986, "case": { "name": "ctx45056-cpu45-f16-b8192-u8192", "context": 45056, "cpu_moe": 45, "kv": "f16", "cache": 0, "batch": 8192, "ubatch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 40550 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C009" ], "source": { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 39305, "sha256": "7eeab3be497b5af10eed75e335d98028459a1845c16df85606ce2d4635989a12" }, "summary_source": { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/summary.json", "bytes": 6332, "sha256": "c6239aa04329a4c8aa48d6348941effda034330d2e6af44f91c00b4211ae4fff" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 212, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21652.0, "board_vram_min_free_mib": 2475.0, "host_ram_min_available_kib": 34623260.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-47104-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T15:21:23.299585+00:00", "ended_utc": "2026-10-05T15:25:21.395536+00:00", "wall_seconds": 238.0959795099916, "case": { "name": "ctx47104-cpu45-f16-b8192-u8192", "context": 47104, "cpu_moe": 45, "kv": "f16", "cache": 0, "batch": 8192, "ubatch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 42393 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C010" ], "source": { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 41392, "sha256": "69db7db6d863e99cb279cfaa8b193e66adb751a440e496a6f101f1582c62dff6" }, "summary_source": { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/summary.json", "bytes": 6334, "sha256": "642fe3113f99462113aa0d3bf4277c3ee2c9158173fe334fc5ba5cde967e7dc7" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 228, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21962.0, "board_vram_min_free_mib": 2165.0, "host_ram_min_available_kib": 34473756.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-49152-cpu46-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T15:07:23.357102+00:00", "ended_utc": "2026-10-05T15:11:19.434181+00:00", "wall_seconds": 236.07711141894106, "case": { "name": "ctx49152-cpu46-f16-b8192-u8192", "context": 49152, "cpu_moe": 46, "kv": "f16", "cache": 0, "batch": 8192, "ubatch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 44236 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C011" ], "source": { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/requests.json", "bytes": 42225, "sha256": "bf6bf41fc09f376f8f2fdf312f21a55d62d1781c6ae80d2f904c31a6662413f5" }, "summary_source": { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/summary.json", "bytes": 6335, "sha256": "d0765d2dc3e2a2b25b973139887d751a99fe0fd75b31efac1f04e4c3a7f5d414" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 226, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21310.0, "board_vram_min_free_mib": 2817.0, "host_ram_min_available_kib": 33344608.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-65536-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T13:50:05.527633+00:00", "ended_utc": "2026-10-05T13:50:05.654546+00:00", "wall_seconds": 0.12693641695659608, "case": { "name": "ctx65536-cpu43-q8_0-b8192", "context": 65536, "cpu_moe": 43, "kv": "q8_0", "cache": 0, "batch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "failure": { "category": "OSError", "message": "[Errno 98] Address already in use" }, "resource_guard_failure": null, "depths": [ 16384, 32768, 49152, 58982 ], "speed_repeats": 3, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [], "source": { "path": "context176-65536-cpu43-q8-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, "summary_source": { "path": "context176-65536-cpu43-q8-20261005/summary.json", "bytes": 1262, "sha256": "b085574e228d2f1c21e3063c97b5c462982ae13117ac1174d7c00061399ecdce" }, "missing_configuration": true, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 1, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 57.0, "board_vram_min_free_mib": 24070.0, "host_ram_min_available_kib": 88922464.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "context176-65536-cpu48-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T15:01:34.346836+00:00", "ended_utc": "2026-10-05T15:06:25.324774+00:00", "wall_seconds": 290.97796507098246, "case": { "name": "ctx65536-cpu48-f16-b8192-u8192", "context": 65536, "cpu_moe": 48, "kv": "f16", "cache": 0, "batch": 8192, "ubatch": 8192, "load_mode": "none", "draft_layers": 99, "n_max": 2, "threads": 14, "thinking": "native default ON" }, "depths": [ 58982 ], "speed_repeats": 1, "quality_repeats": 1, "native_model_max_context": 262144, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C012" ], "source": { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/requests.json", "bytes": 50549, "sha256": "dc983c4ef8356d366d761ce45d9b5e990f887f0c425068b76374a43b05ff494c" }, "summary_source": { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/summary.json", "bytes": 6334, "sha256": "c0e45ce846e94547e8e8f0fff83e33ea68d34fd28b2c554122d3cddd12c3e5dc" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 279, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21870.0, "board_vram_min_free_mib": 2257.0, "host_ram_min_available_kib": 29821048.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T10:16:20.399039+00:00", "ended_utc": "2026-10-05T10:21:53.703126+00:00", "wall_seconds": 333.3041146880714, "thinking": "native default ON", "speed_repeats": 1, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C013", "C014", "C015", "C016" ], "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "summary_source": { "path": "guided-speed175-20261005/summary.json", "bytes": 14075, "sha256": "ea42c156b8ccbf351b9b057c8542a4390366b7cb982b9cc4d98339f32625afff" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 339, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21072.0, "board_vram_min_free_mib": 3055.0, "host_ram_min_available_kib": 83653876.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T11:21:13.939131+00:00", "ended_utc": "2026-10-05T11:27:24.311594+00:00", "wall_seconds": 370.3724879089277, "thinking": "native default ON", "speed_repeats": 5, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C017" ], "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "summary_source": { "path": "guided-speed175-finalist-20261005/summary.json", "bytes": 7727, "sha256": "285e6cd2c9ddc912cfef0d28e59b93c2bda9a05419ee22bff25e6b5407404a33" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 373, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18718.0, "board_vram_min_free_mib": 5409.0, "host_ram_min_available_kib": 83506092.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T10:41:41.977432+00:00", "ended_utc": "2026-10-05T10:54:55.733764+00:00", "wall_seconds": 793.7563566310564, "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C018", "C019", "C020" ], "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "summary_source": { "path": "guided-speed175-followup-20261005/summary.json", "bytes": 12706, "sha256": "aa8a052a4d5da522dfc392a06c85a8b7e242ba2b9e74ec6766f0b0e6dbe0ed13" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 798, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 20342.0, "board_vram_min_free_mib": 3785.0, "host_ram_min_available_kib": 79720432.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T11:03:55.182904+00:00", "ended_utc": "2026-10-05T11:18:25.367064+00:00", "wall_seconds": 870.1841906160116, "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C021", "C022", "C023" ], "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "summary_source": { "path": "guided-speed175-native-profile-20261005/summary.json", "bytes": 11914, "sha256": "e123af9cdcca4346e839c21d3b30755f9087429d68f54124190ca3ac28fd3c20" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 874, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 19942.0, "board_vram_min_free_mib": 4185.0, "host_ram_min_available_kib": 82068540.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "historical180/mtp-screen32-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_MATRIX_ABORTED_NO_RETRY", "summary": { "status": "FAIL_MATRIX_ABORTED_NO_RETRY", "started_utc": "2026-10-05T07:42:20.464590+00:00", "ended_utc": "2026-10-05T07:43:32.991494+00:00", "wall_seconds": 72.52693182602525, "failure": { "type": "QAError", "message": "native expert-cache64 engagement not proved" } }, "configuration_ids": [ "C024" ], "source": { "path": "historical180/mtp-screen32-20261005/requests.jsonl", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, "summary_source": { "path": "historical180/mtp-screen32-20261005/summary.json", "bytes": 24730, "sha256": "c11095ce7e544e9351ddffee024a3262fec8b7cbabca65bfa0f88cbb23a5d8bf" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf", "bytes": 10946624, "sha256": "4448186216b3af4cc558bbce2c3213f01608f8f8b2e5267a9767971dd3ec8082" }, { "path": "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00002-of-00004.gguf", "bytes": 49859583136, "sha256": "3f342f1c1580473f1ee94ddd5b28206e8c07a70fa1a366f59d1d6c922919a6c9" }, { "path": "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00003-of-00004.gguf", "bytes": 49376141504, "sha256": "56758f40269cad5cd9b0d3d6fbae0f40f6d5be6de49e4ab392dbe83157d9cbd3" }, { "path": "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00004-of-00004.gguf", "bytes": 12087983520, "sha256": "753bda48b98ba4f1636134a90a967de1b2d3908a236c026e464777342e53510a" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 73, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 17778.0, "board_vram_min_free_mib": 6349.0, "host_ram_min_available_kib": 87182568.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "NO_SUMMARY_STATUS", "summary": {}, "configuration_ids": [ "C025" ], "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 0, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": null, "board_vram_min_free_mib": null, "host_ram_min_available_kib": null, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "NO_SUMMARY_STATUS", "summary": {}, "configuration_ids": [ "C026" ], "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 0, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": null, "board_vram_min_free_mib": null, "host_ram_min_available_kib": null, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "NO_SUMMARY_STATUS", "summary": {}, "configuration_ids": [ "C027" ], "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 0, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": null, "board_vram_min_free_mib": null, "host_ram_min_available_kib": null, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "NO_SUMMARY_STATUS", "summary": {}, "configuration_ids": [ "C028" ], "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 0, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": null, "board_vram_min_free_mib": null, "host_ram_min_available_kib": null, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "ik178/results/control32-noMTP", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T19:01:01.882934+00:00", "ended_utc": "2026-10-05T19:04:25.109574+00:00", "wall_seconds": 203.2266597050475, "thinking": "native ON", "text_only": true, "engine_commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C029" ], "source": { "path": "ik178/results/control32-noMTP/requests.json", "bytes": 36758, "sha256": "ca40d5e5b17d4f6f759460c4acdac5fbd606fb28cbbd3acb8284d11c7053356e" }, "summary_source": { "path": "ik178/results/control32-noMTP/summary.json", "bytes": 5586, "sha256": "9b9be58d9282469e9c2aba8149c325b6a58226fe8d4ed47a845edb502d3be76c" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 207, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18950.0, "board_vram_min_free_mib": 5177.0, "host_ram_min_available_kib": 41576384.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "ik178/results/mtp32-n1", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T19:07:16.395976+00:00", "ended_utc": "2026-10-05T19:08:40.931883+00:00", "wall_seconds": 84.53595128306188, "thinking": "native ON", "text_only": true, "engine_commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C030" ], "source": { "path": "ik178/results/mtp32-n1/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, "summary_source": { "path": "ik178/results/mtp32-n1/summary.json", "bytes": 5169, "sha256": "3a806194e6fb5f40482c99875e9d4515b995d654195f487ffff2c8e3c82fedba" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 83, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21416.0, "board_vram_min_free_mib": 2711.0, "host_ram_min_available_kib": 42553540.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "ik178/results/mtp32-n1-draftU512", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T19:11:51.671047+00:00", "ended_utc": "2026-10-05T19:13:19.682148+00:00", "wall_seconds": 88.01115805399604, "thinking": "native ON", "text_only": true, "engine_commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C031" ], "source": { "path": "ik178/results/mtp32-n1-draftU512/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, "summary_source": { "path": "ik178/results/mtp32-n1-draftU512/summary.json", "bytes": 5590, "sha256": "6136b95fc0c866ba4e61fbf3487405d6ee451b17ea050a65a16c10a5ddcc7bcc" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 87, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 22200.0, "board_vram_min_free_mib": 1927.0, "host_ram_min_available_kib": 34173776.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "ik178/results/mtp32-n1-mainU4096-draftU512", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T19:14:25.495741+00:00", "ended_utc": "2026-10-05T19:18:02.735719+00:00", "wall_seconds": 217.24000533402432, "thinking": "native ON", "text_only": true, "engine_commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C032" ], "source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.json", "bytes": 40991, "sha256": "b0c4b235b06560ccf360509232909a39c3a61c50623d801034a31ca09e12f274" }, "summary_source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/summary.json", "bytes": 6255, "sha256": "37c8d10eb66e50bd0774a19f93ab523e39249403d4c38acdafd0988ebff3986d" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 221, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18528.0, "board_vram_min_free_mib": 5599.0, "host_ram_min_available_kib": 24812716.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T08:40:22.587955+00:00", "ended_utc": "2026-10-05T08:55:57.187864+00:00", "wall_seconds": 934.5999433220131, "thinking": "native default", "speed_repeats": 3, "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C033", "C034" ], "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "summary_source": { "path": "iq3-author32-accept-20261005/summary.json", "bytes": 9686, "sha256": "194b8d985220e729963077558d0fc8c195ce7b088bad39f6a6bfb5b8360cbf5d" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 938, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10804.0, "board_vram_min_free_mib": 13323.0, "host_ram_min_available_kib": 84513132.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T09:00:18.629783+00:00", "ended_utc": "2026-10-05T09:07:39.067793+00:00", "wall_seconds": 440.43803767696954, "thinking": "native default", "speed_repeats": 3, "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C035" ], "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "summary_source": { "path": "iq3-cache96-accept-20261005/summary.json", "bytes": 7359, "sha256": "83c61070b3e2220f45254a8aac6a57b4a7b3ec6fb22a7b6f3261ae45df3f3399" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 443, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 14820.0, "board_vram_min_free_mib": 9307.0, "host_ram_min_available_kib": 85385860.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T09:12:13.601069+00:00", "ended_utc": "2026-10-05T09:16:28.251696+00:00", "wall_seconds": 254.65064873197116, "thinking": "native default", "speed_repeats": 3, "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C036" ], "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "summary_source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/summary.json", "bytes": 7392, "sha256": "922f2a15c82bd1e27fe1ff871f1cb2a201c907b607ca767af279fcfd2cbfebe9" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 257, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16986.0, "board_vram_min_free_mib": 7141.0, "host_ram_min_available_kib": 84759420.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "opt181/results/opt256-cache4g-ple-overlap-workspace", "engine": "OptLlama", "commit": "167742d9dfa01306786e7d3cfd31fcb8a8b185dc", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T20:30:17.700421+00:00", "ended_utc": "2026-10-05T20:33:45.936619+00:00", "wall_seconds": 208.23623241099995, "case": { "name": "opt256-cache4g-ple-overlap-workspace", "context": 262144, "cpu_moe": 48, "kv": "f16", "batch": 8192, "ubatch": 4096, "mtp": 0, "variant": "overlap", "phase_live_workspace": true, "cache_budget_mib": 4096, "expert_placement": "native CUDA cache claims routed tensors" }, "thinking": "native ON", "depths": [ 8192, 235929 ], "text_only": true }, "configuration_ids": [ "C037" ], "source": { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/requests.json", "bytes": 14071, "sha256": "4c0b93e6ac36788f660eaf6e72ec5fbdc2ad3baa70ac79e3b1d8fc0f06eb8b06" }, "summary_source": { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/summary.json", "bytes": 7148, "sha256": "d81e16005e2f64a28c0d74ed8db3ed02166a0840fd0895471a30b9ea7f3b9d4c" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ] }, "resources": { "sample_count": 167, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 23930.0, "board_vram_min_free_mib": 197.0, "host_ram_min_available_kib": 37482796.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T11:54:38.641975+00:00", "ended_utc": "2026-10-05T11:59:21.034331+00:00", "wall_seconds": 282.3923893540632, "failure": { "category": "ConnectionResetError", "message": "[Errno 104] Connection reset by peer" }, "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C038", "C039" ], "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "summary_source": { "path": "placement-speed175-20261005/summary.json", "bytes": 9845, "sha256": "97b2af536849938c2ea85ebc0736e30b31a85c104e39ebdca4ec2c9ccfc68e71" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 283, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18718.0, "board_vram_min_free_mib": 5409.0, "host_ram_min_available_kib": 83465880.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "placement-speed175-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T12:04:52.555174+00:00", "ended_utc": "2026-10-05T12:05:47.283911+00:00", "wall_seconds": 54.72879904892761, "failure": { "category": "QAError", "message": "TimeoutExpired: Command '['/usr/bin/nvidia-smi', '--query-compute-apps=pid', '--format=csv,noheader,nounits']' timed out after 8 seconds" }, "resource_guard_failure": "TimeoutExpired: Command '['/usr/bin/nvidia-smi', '--query-compute-apps=pid', '--format=csv,noheader,nounits']' timed out after 8 seconds", "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C040" ], "source": { "path": "placement-speed175-r1-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, "summary_source": { "path": "placement-speed175-r1-20261005/summary.json", "bytes": 8148, "sha256": "96d565892876f02a3ead7c23ec4009573372d2ca6119a2ce24e386a655cc96fd" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 44, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 318.0, "board_vram_min_free_mib": 23809.0, "host_ram_min_available_kib": 87944396.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T12:08:03.865611+00:00", "ended_utc": "2026-10-05T12:13:34.958902+00:00", "wall_seconds": 331.0933254819829, "failure": { "category": "QAError", "message": "QA server did not stop gracefully" }, "resource_guard_failure": "TimeoutExpired: Command '['/usr/bin/nvidia-smi', '--query-compute-apps=pid', '--format=csv,noheader,nounits']' timed out after 0.7 seconds", "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C041" ], "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "summary_source": { "path": "placement-speed175-r2-20261005/summary.json", "bytes": 8078, "sha256": "e6fcfca339f3f477f957f55822dceec941db3206468c6ebf25f0ed69308fb74e" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 308, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18810.0, "board_vram_min_free_mib": 5317.0, "host_ram_min_available_kib": 35121848.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T12:36:10.750749+00:00", "ended_utc": "2026-10-05T12:40:29.964665+00:00", "wall_seconds": 259.21393788198475, "failure": { "category": "QAError", "message": "TimeoutExpired: Command '['/usr/bin/nvidia-smi', '--query-compute-apps=pid', '--format=csv,noheader,nounits']' timed out after 0.7 seconds" }, "resource_guard_failure": "TimeoutExpired: Command '['/usr/bin/nvidia-smi', '--query-compute-apps=pid', '--format=csv,noheader,nounits']' timed out after 0.7 seconds", "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C042" ], "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "summary_source": { "path": "prefill-dma175-20261005/summary.json", "bytes": 11354, "sha256": "b6b9ba5307cc1016156bfb4cc0767c9d1fdcb36af3c980583b1fc77fc7de6d81" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 244, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 22050.0, "board_vram_min_free_mib": 2077.0, "host_ram_min_available_kib": 38337576.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T12:55:35.407377+00:00", "ended_utc": "2026-10-05T13:06:28.260789+00:00", "wall_seconds": 652.8534393600421, "thinking": "native default ON", "speed_repeats": 5, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C043", "C044" ], "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "summary_source": { "path": "prefill-finalist175-20261005/summary.json", "bytes": 11422, "sha256": "3ca2fda59b6b96253c509771c38e0406e21859326a542c51640d7f71172cf3af" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 634, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21582.0, "board_vram_min_free_mib": 2545.0, "host_ram_min_available_kib": 36743316.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T12:22:25.131812+00:00", "ended_utc": "2026-10-05T12:29:32.390382+00:00", "wall_seconds": 427.25859917805064, "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C045", "C046" ], "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "summary_source": { "path": "prefill-hybrid175-20261005/summary.json", "bytes": 10507, "sha256": "f0fefde7811f4a123a12a13d0b9a4aec6c8a5a30836d4f34547d6cc30a36057c" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 433, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21970.0, "board_vram_min_free_mib": 2157.0, "host_ram_min_available_kib": 80802036.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T12:45:24.561748+00:00", "ended_utc": "2026-10-05T12:54:20.641484+00:00", "wall_seconds": 536.0797763379524, "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C047", "C048" ], "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "summary_source": { "path": "prefill-kv175-20261005/summary.json", "bytes": 11420, "sha256": "38c17fbfad6e801ca62438be32f0631da68d601dfc23cea5cb5b6ccddbf705d8" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 517, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21700.0, "board_vram_min_free_mib": 2427.0, "host_ram_min_available_kib": 37002176.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T13:11:53.678848+00:00", "ended_utc": "2026-10-05T13:17:49.291437+00:00", "wall_seconds": 355.6126215480035, "thinking": "native default ON", "speed_repeats": 5, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C049" ], "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "summary_source": { "path": "prefill-mtp1-finalist175-20261005/summary.json", "bytes": 8570, "sha256": "60c449a49076ec96ec75a1567092629cce343d61e0f421f4d616aa3d9c920863" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 347, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21470.0, "board_vram_min_free_mib": 2657.0, "host_ram_min_available_kib": 36986416.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-observe175-20261005/benchmark", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "NO_SUMMARY_STATUS", "summary": {}, "configuration_ids": [], "source": { "path": "prefill-observe175-20261005/benchmark/context-32768.jsonl", "bytes": 12459, "sha256": "2ba68e02684518163fcf67ff0d66d96a5d620a8a4440facf5bf32a0703ac3e17" }, "summary_source": null, "missing_configuration": true, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 0, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": null, "board_vram_min_free_mib": null, "host_ram_min_available_kib": null, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T12:17:24.885586+00:00", "ended_utc": "2026-10-05T12:21:00.235045+00:00", "wall_seconds": 215.3494882209925, "thinking": "native default ON", "speed_repeats": 3, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "model_revision": "38bb39ee97821de2c9009abb7e93950eec396e66" }, "configuration_ids": [ "C050" ], "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "summary_source": { "path": "prefill-path175-20261005/summary.json", "bytes": 7772, "sha256": "90e3225f94522e357d6f5d294b06ba284f2dc3b5688076ff24be36c6fd5fa7c6" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 219, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10880.0, "board_vram_min_free_mib": 13247.0, "host_ram_min_available_kib": 84952568.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/control32-noMTP", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T17:36:38.248351+00:00", "ended_utc": "2026-10-05T17:40:02.482097+00:00", "wall_seconds": 204.23376805195585, "thinking": "native ON", "depths": [ 8192, 28672 ], "text_only": true }, "configuration_ids": [ "C051" ], "source": { "path": "upstream178/results/control32-noMTP/requests.json", "bytes": 34778, "sha256": "41ee43a7a40ec41fd35382ef0280d587483484892350d003f9d4072c74cd3db0" }, "summary_source": { "path": "upstream178/results/control32-noMTP/summary.json", "bytes": 6496, "sha256": "5c0debeab732d308cd55543ee0d9dd135d76dde4915bbcd1eedf0b36f47cb23f" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "manifest_sha256": "b3dbf71157008f67943d3204f3d5caebfa981b030e917c8ba8ce2990a84a8d8a", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" }, { "path": "/llama-server", "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" } ] } }, "resources": { "sample_count": 195, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16268.0, "board_vram_min_free_mib": 7859.0, "host_ram_min_available_kib": 41335116.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T19:18:24.947024+00:00", "ended_utc": "2026-10-05T19:38:58.639101+00:00", "wall_seconds": 1233.6921035420382, "thinking": "native ON", "depths": [ 8192, 131072, 235929 ], "text_only": true }, "configuration_ids": [ "C052" ], "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "summary_source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/summary.json", "bytes": 6930, "sha256": "46a8cfacb8e8799cc45e0c8a11a29d64ebef40ef180086b84f77ea781d07df2c" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "8345f333951c661d166b00e6f9362e553768f292", "manifest_sha256": "444e536dca0e51390e74c79c24805c573c35883cd8bd53ae1d80d9439d4febda", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.26.0", "sha256": "a64e2d094cab61f46894d5310ce7dce9688423de843532ee05e1b5189246bd66" }, { "path": "/libggml-cpu.so.0.26.0", "sha256": "5db3086ad649c894f487104325bf24588ee28914d79a1e6b74ddd53df727aeb5" }, { "path": "/libggml-cuda.so.0.26.0", "sha256": "ae3c9766d202a8a5946c93937c532906a7db46f04283a4a751fb0e1139b0212e" }, { "path": "/libggml.so.0.26.0", "sha256": "542219e3a0a7c7bc8b1515852d68c051cdea3f866e7af4f8f275ed387a5d80ed" }, { "path": "/libllama-bench-impl.so", "sha256": "d48211bd07a2e1eeddb6fd0fd6a186f1839afd87643a678414b61b40ed9360df" }, { "path": "/libllama-common.so.0.6.0", "sha256": "3fa61a74da7f585d3de6d93340e08477d4561e3f696bfe4109f39b4de30c9dce" }, { "path": "/libllama-server-impl.so", "sha256": "2d6c5457b5b461b709a536b26f366d5a0c499c0d5917c4e486c3e25532f97519" }, { "path": "/libllama.so.0.6.0", "sha256": "9b510d5c9d6ca8d361045e12a5c6e074697e33f4270da1bf500498e81fa63586" }, { "path": "/libmtmd.so.0.6.0", "sha256": "bd1ba5348fb2cc64850efe2fd69eb83ba0f7eceb3f0c71e5677fab5f757d2df3" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "c368acfcf10e5e1732e50e5364fbeb7b716a7ae0c07d13366db6f808bcf2a969" }, { "path": "/llama-server", "sha256": "24c8788d3af1bf22f7aa6e0166247d16c70b75091f8326c8d0e7dd1f735d8794" } ] } }, "resources": { "sample_count": 1223, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21338.0, "board_vram_min_free_mib": 2789.0, "host_ram_min_available_kib": 34565076.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/upstream256-cpu46-u4096-deep179", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T19:41:51.777941+00:00", "ended_utc": "2026-10-05T19:52:37.386691+00:00", "wall_seconds": 645.6087789019803, "thinking": "native ON", "depths": [ 235929 ], "text_only": true }, "configuration_ids": [ "C053" ], "source": { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/requests.json", "bytes": 120120, "sha256": "3b7049b12e032ae91f9608158dee9267dcb16187833d5bfbb045f6b123421a8b" }, "summary_source": { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/summary.json", "bytes": 6373, "sha256": "364bfb1390edb821e2b87e1921af4f340f17d0c7f291d447b793c03437bbdf27" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "8345f333951c661d166b00e6f9362e553768f292", "manifest_sha256": "444e536dca0e51390e74c79c24805c573c35883cd8bd53ae1d80d9439d4febda", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.26.0", "sha256": "a64e2d094cab61f46894d5310ce7dce9688423de843532ee05e1b5189246bd66" }, { "path": "/libggml-cpu.so.0.26.0", "sha256": "5db3086ad649c894f487104325bf24588ee28914d79a1e6b74ddd53df727aeb5" }, { "path": "/libggml-cuda.so.0.26.0", "sha256": "ae3c9766d202a8a5946c93937c532906a7db46f04283a4a751fb0e1139b0212e" }, { "path": "/libggml.so.0.26.0", "sha256": "542219e3a0a7c7bc8b1515852d68c051cdea3f866e7af4f8f275ed387a5d80ed" }, { "path": "/libllama-bench-impl.so", "sha256": "d48211bd07a2e1eeddb6fd0fd6a186f1839afd87643a678414b61b40ed9360df" }, { "path": "/libllama-common.so.0.6.0", "sha256": "3fa61a74da7f585d3de6d93340e08477d4561e3f696bfe4109f39b4de30c9dce" }, { "path": "/libllama-server-impl.so", "sha256": "2d6c5457b5b461b709a536b26f366d5a0c499c0d5917c4e486c3e25532f97519" }, { "path": "/libllama.so.0.6.0", "sha256": "9b510d5c9d6ca8d361045e12a5c6e074697e33f4270da1bf500498e81fa63586" }, { "path": "/libmtmd.so.0.6.0", "sha256": "bd1ba5348fb2cc64850efe2fd69eb83ba0f7eceb3f0c71e5677fab5f757d2df3" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "c368acfcf10e5e1732e50e5364fbeb7b716a7ae0c07d13366db6f808bcf2a969" }, { "path": "/llama-server", "sha256": "24c8788d3af1bf22f7aa6e0166247d16c70b75091f8326c8d0e7dd1f735d8794" } ] } }, "resources": { "sample_count": 634, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 22050.0, "board_vram_min_free_mib": 2077.0, "host_ram_min_available_kib": 30190664.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/upstream256-cpu46-u4096-short179", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T19:39:18.478941+00:00", "ended_utc": "2026-10-05T19:41:29.351965+00:00", "wall_seconds": 130.87305558403023, "thinking": "native ON", "depths": [ 8192 ], "text_only": true }, "configuration_ids": [ "C054" ], "source": { "path": "upstream178/results/upstream256-cpu46-u4096-short179/requests.json", "bytes": 14476, "sha256": "c10d3cfc6128e73ddea302b574fd15edd45492afe5b368cc83d99703d23ab7e3" }, "summary_source": { "path": "upstream178/results/upstream256-cpu46-u4096-short179/summary.json", "bytes": 6367, "sha256": "9b0adac2f2dbff4be04fddef2aa90b094caafd2f1c217a6711956c6453044ef0" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "8345f333951c661d166b00e6f9362e553768f292", "manifest_sha256": "444e536dca0e51390e74c79c24805c573c35883cd8bd53ae1d80d9439d4febda", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.26.0", "sha256": "a64e2d094cab61f46894d5310ce7dce9688423de843532ee05e1b5189246bd66" }, { "path": "/libggml-cpu.so.0.26.0", "sha256": "5db3086ad649c894f487104325bf24588ee28914d79a1e6b74ddd53df727aeb5" }, { "path": "/libggml-cuda.so.0.26.0", "sha256": "ae3c9766d202a8a5946c93937c532906a7db46f04283a4a751fb0e1139b0212e" }, { "path": "/libggml.so.0.26.0", "sha256": "542219e3a0a7c7bc8b1515852d68c051cdea3f866e7af4f8f275ed387a5d80ed" }, { "path": "/libllama-bench-impl.so", "sha256": "d48211bd07a2e1eeddb6fd0fd6a186f1839afd87643a678414b61b40ed9360df" }, { "path": "/libllama-common.so.0.6.0", "sha256": "3fa61a74da7f585d3de6d93340e08477d4561e3f696bfe4109f39b4de30c9dce" }, { "path": "/libllama-server-impl.so", "sha256": "2d6c5457b5b461b709a536b26f366d5a0c499c0d5917c4e486c3e25532f97519" }, { "path": "/libllama.so.0.6.0", "sha256": "9b510d5c9d6ca8d361045e12a5c6e074697e33f4270da1bf500498e81fa63586" }, { "path": "/libmtmd.so.0.6.0", "sha256": "bd1ba5348fb2cc64850efe2fd69eb83ba0f7eceb3f0c71e5677fab5f757d2df3" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "c368acfcf10e5e1732e50e5364fbeb7b716a7ae0c07d13366db6f808bcf2a969" }, { "path": "/llama-server", "sha256": "24c8788d3af1bf22f7aa6e0166247d16c70b75091f8326c8d0e7dd1f735d8794" } ] } }, "resources": { "sample_count": 119, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21858.0, "board_vram_min_free_mib": 2269.0, "host_ram_min_available_kib": 37426268.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T18:02:55.887275+00:00", "ended_utc": "2026-10-05T18:19:37.962948+00:00", "wall_seconds": 1002.0757038920419, "thinking": "native ON", "depths": [ 8192, 131072, 235929 ], "text_only": true }, "configuration_ids": [ "C055" ], "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "summary_source": { "path": "upstream178/results/upstream256-f16-noMTP/summary.json", "bytes": 6932, "sha256": "c999cba80d5ade26cfb4b9e93f63295854dd7118644150a8cc7967b10c881c6a" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "8345f333951c661d166b00e6f9362e553768f292", "manifest_sha256": "444e536dca0e51390e74c79c24805c573c35883cd8bd53ae1d80d9439d4febda", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.26.0", "sha256": "a64e2d094cab61f46894d5310ce7dce9688423de843532ee05e1b5189246bd66" }, { "path": "/libggml-cpu.so.0.26.0", "sha256": "5db3086ad649c894f487104325bf24588ee28914d79a1e6b74ddd53df727aeb5" }, { "path": "/libggml-cuda.so.0.26.0", "sha256": "ae3c9766d202a8a5946c93937c532906a7db46f04283a4a751fb0e1139b0212e" }, { "path": "/libggml.so.0.26.0", "sha256": "542219e3a0a7c7bc8b1515852d68c051cdea3f866e7af4f8f275ed387a5d80ed" }, { "path": "/libllama-bench-impl.so", "sha256": "d48211bd07a2e1eeddb6fd0fd6a186f1839afd87643a678414b61b40ed9360df" }, { "path": "/libllama-common.so.0.6.0", "sha256": "3fa61a74da7f585d3de6d93340e08477d4561e3f696bfe4109f39b4de30c9dce" }, { "path": "/libllama-server-impl.so", "sha256": "2d6c5457b5b461b709a536b26f366d5a0c499c0d5917c4e486c3e25532f97519" }, { "path": "/libllama.so.0.6.0", "sha256": "9b510d5c9d6ca8d361045e12a5c6e074697e33f4270da1bf500498e81fa63586" }, { "path": "/libmtmd.so.0.6.0", "sha256": "bd1ba5348fb2cc64850efe2fd69eb83ba0f7eceb3f0c71e5677fab5f757d2df3" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "c368acfcf10e5e1732e50e5364fbeb7b716a7ae0c07d13366db6f808bcf2a969" }, { "path": "/llama-server", "sha256": "24c8788d3af1bf22f7aa6e0166247d16c70b75091f8326c8d0e7dd1f735d8794" } ] } }, "resources": { "sample_count": 991, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 20128.0, "board_vram_min_free_mib": 3999.0, "host_ram_min_available_kib": 28090068.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T17:45:37.016691+00:00", "ended_utc": "2026-10-05T18:02:40.511617+00:00", "wall_seconds": 1023.4949477401096, "thinking": "native ON", "depths": [ 8192, 131072, 235929 ], "text_only": true }, "configuration_ids": [ "C056" ], "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "summary_source": { "path": "upstream178/results/upstream256-noMTP/summary.json", "bytes": 6933, "sha256": "d6f234b71ffcd674d84abb60a8e9585f00628fe063ee50b2d449d661108e431b" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "8345f333951c661d166b00e6f9362e553768f292", "manifest_sha256": "444e536dca0e51390e74c79c24805c573c35883cd8bd53ae1d80d9439d4febda", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.26.0", "sha256": "a64e2d094cab61f46894d5310ce7dce9688423de843532ee05e1b5189246bd66" }, { "path": "/libggml-cpu.so.0.26.0", "sha256": "5db3086ad649c894f487104325bf24588ee28914d79a1e6b74ddd53df727aeb5" }, { "path": "/libggml-cuda.so.0.26.0", "sha256": "ae3c9766d202a8a5946c93937c532906a7db46f04283a4a751fb0e1139b0212e" }, { "path": "/libggml.so.0.26.0", "sha256": "542219e3a0a7c7bc8b1515852d68c051cdea3f866e7af4f8f275ed387a5d80ed" }, { "path": "/libllama-bench-impl.so", "sha256": "d48211bd07a2e1eeddb6fd0fd6a186f1839afd87643a678414b61b40ed9360df" }, { "path": "/libllama-common.so.0.6.0", "sha256": "3fa61a74da7f585d3de6d93340e08477d4561e3f696bfe4109f39b4de30c9dce" }, { "path": "/libllama-server-impl.so", "sha256": "2d6c5457b5b461b709a536b26f366d5a0c499c0d5917c4e486c3e25532f97519" }, { "path": "/libllama.so.0.6.0", "sha256": "9b510d5c9d6ca8d361045e12a5c6e074697e33f4270da1bf500498e81fa63586" }, { "path": "/libmtmd.so.0.6.0", "sha256": "bd1ba5348fb2cc64850efe2fd69eb83ba0f7eceb3f0c71e5677fab5f757d2df3" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "c368acfcf10e5e1732e50e5364fbeb7b716a7ae0c07d13366db6f808bcf2a969" }, { "path": "/llama-server", "sha256": "24c8788d3af1bf22f7aa6e0166247d16c70b75091f8326c8d0e7dd1f735d8794" } ] } }, "resources": { "sample_count": 1012, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16532.0, "board_vram_min_free_mib": 7595.0, "host_ram_min_available_kib": 29012528.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "upstream178/results/upstream32-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "server-requests", "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "summary": { "status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "started_utc": "2026-10-05T17:40:54.632821+00:00", "ended_utc": "2026-10-05T17:44:11.243002+00:00", "wall_seconds": 196.61021151195746, "thinking": "native ON", "depths": [ 8192, 28672 ], "text_only": true }, "configuration_ids": [ "C057" ], "source": { "path": "upstream178/results/upstream32-noMTP/requests.json", "bytes": 32699, "sha256": "2e478a9a8ef727b7a0d4966bcfde2dd6b9b62df6a4c61de2344ed8ffef55b0f6" }, "summary_source": { "path": "upstream178/results/upstream32-noMTP/summary.json", "bytes": 6642, "sha256": "2bdd58a9329b42a8d1800349601c64370732f230d6809f4622dd91d5a33bf7bc" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "model_integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" } ], "engine": { "commit": "8345f333951c661d166b00e6f9362e553768f292", "manifest_sha256": "444e536dca0e51390e74c79c24805c573c35883cd8bd53ae1d80d9439d4febda", "artifacts": [ { "path": "/libcublas.so.12", "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.26.0", "sha256": "a64e2d094cab61f46894d5310ce7dce9688423de843532ee05e1b5189246bd66" }, { "path": "/libggml-cpu.so.0.26.0", "sha256": "5db3086ad649c894f487104325bf24588ee28914d79a1e6b74ddd53df727aeb5" }, { "path": "/libggml-cuda.so.0.26.0", "sha256": "ae3c9766d202a8a5946c93937c532906a7db46f04283a4a751fb0e1139b0212e" }, { "path": "/libggml.so.0.26.0", "sha256": "542219e3a0a7c7bc8b1515852d68c051cdea3f866e7af4f8f275ed387a5d80ed" }, { "path": "/libllama-bench-impl.so", "sha256": "d48211bd07a2e1eeddb6fd0fd6a186f1839afd87643a678414b61b40ed9360df" }, { "path": "/libllama-common.so.0.6.0", "sha256": "3fa61a74da7f585d3de6d93340e08477d4561e3f696bfe4109f39b4de30c9dce" }, { "path": "/libllama-server-impl.so", "sha256": "2d6c5457b5b461b709a536b26f366d5a0c499c0d5917c4e486c3e25532f97519" }, { "path": "/libllama.so.0.6.0", "sha256": "9b510d5c9d6ca8d361045e12a5c6e074697e33f4270da1bf500498e81fa63586" }, { "path": "/libmtmd.so.0.6.0", "sha256": "bd1ba5348fb2cc64850efe2fd69eb83ba0f7eceb3f0c71e5677fab5f757d2df3" }, { "path": "/libnccl.so.2", "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "sha256": "c368acfcf10e5e1732e50e5364fbeb7b716a7ae0c07d13366db6f808bcf2a969" }, { "path": "/llama-server", "sha256": "24c8788d3af1bf22f7aa6e0166247d16c70b75091f8326c8d0e7dd1f735d8794" } ] } }, "resources": { "sample_count": 187, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 14732.0, "board_vram_min_free_mib": 9395.0, "host_ram_min_available_kib": 41273288.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C058" ], "source": { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/results.jsonl", "bytes": 2265, "sha256": "7c7c17e0cd1fa69e131cff29b73a6cb3406e3155080768b3e512fba00b9c67f0" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 27, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10760.0, "board_vram_min_free_mib": 13367.0, "host_ram_min_available_kib": 87876456.0, "process_rss_peak_kib": 75736004.0, "cgroup_peak_bytes_sampled": 858062848.0 } }, { "id": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C059" ], "source": { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/results.jsonl", "bytes": 2264, "sha256": "36dd6008ff9c726d1ef1bf218390853667493641c8c0eb89497710de9c556020" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 26, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 11908.0, "board_vram_min_free_mib": 12219.0, "host_ram_min_available_kib": 87996808.0, "process_rss_peak_kib": 75809004.0, "cgroup_peak_bytes_sampled": 858669056.0 } }, { "id": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C060" ], "source": { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/results.jsonl", "bytes": 2265, "sha256": "9a8a4e39214934adff34bc4755b8cdd4ac245f4cd37f3345f795ac4cdc62d2cd" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 26, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 13056.0, "board_vram_min_free_mib": 11071.0, "host_ram_min_available_kib": 87897776.0, "process_rss_peak_kib": 75887208.0, "cgroup_peak_bytes_sampled": 857174016.0 } }, { "id": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C061" ], "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/results.jsonl", "bytes": 2264, "sha256": "51b9db09d6f4c913e4f86a2833a7cddd68cb355e25c9dfd33c8da79ec4faa68f" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 24, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 13056.0, "board_vram_min_free_mib": 11071.0, "host_ram_min_available_kib": 87930776.0, "process_rss_peak_kib": 75929092.0, "cgroup_peak_bytes_sampled": 899674112.0 } }, { "id": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C062" ], "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/results.jsonl", "bytes": 2263, "sha256": "dbca6964ce8a5ce3149b84b5c70adb33b614ea59c6e36968f5893a6a02a5c258" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 26, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 13056.0, "board_vram_min_free_mib": 11071.0, "host_ram_min_available_kib": 87922484.0, "process_rss_peak_kib": 75886376.0, "cgroup_peak_bytes_sampled": 856305664.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C063" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/results.jsonl", "bytes": 1137, "sha256": "8889121a50e4975cfc60c415e21f5c8619476775f72a6bf05e21d9af74e724e8" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 28, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 11108.0, "board_vram_min_free_mib": 13019.0, "host_ram_min_available_kib": 87854252.0, "process_rss_peak_kib": 78735464.0, "cgroup_peak_bytes_sampled": 889348096.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C064" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/results.jsonl", "bytes": 1136, "sha256": "fa27ba388a0d88ac40739baaea1796a11f01d6083a0b3b9a15973f35f169776d" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 96, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10676.0, "board_vram_min_free_mib": 13451.0, "host_ram_min_available_kib": 87854628.0, "process_rss_peak_kib": 78651940.0, "cgroup_peak_bytes_sampled": 850092032.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C065" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/results.jsonl", "bytes": 1136, "sha256": "5a7e4df43f0b0b56f751acb431f15e6498d5e7eb27a69bd78ba1ae30e0a3d00e" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 63, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10738.0, "board_vram_min_free_mib": 13389.0, "host_ram_min_available_kib": 87876916.0, "process_rss_peak_kib": 78661524.0, "cgroup_peak_bytes_sampled": 853782528.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C066" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/results.jsonl", "bytes": 1137, "sha256": "39888789994ee079760655a6a7003a4e91b57ab865e7bb65b9863bda3b947afc" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 41, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10860.0, "board_vram_min_free_mib": 13267.0, "host_ram_min_available_kib": 87881424.0, "process_rss_peak_kib": 78698800.0, "cgroup_peak_bytes_sampled": 866430976.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C067" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/results.jsonl", "bytes": 1137, "sha256": "661868b3ce91d08f020e3bae830898d6045fcd588250416e22bbb1ec1d04881c" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 28, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 11108.0, "board_vram_min_free_mib": 13019.0, "host_ram_min_available_kib": 87864424.0, "process_rss_peak_kib": 78736808.0, "cgroup_peak_bytes_sampled": 891281408.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "BENCH_OR_MONITOR_FAILED", "summary": {}, "configuration_ids": [ "C068" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "bytes": 3221, "sha256": "93736ea0d0dd4828ceb3da78311eac90b6ab9e0bbc581e72ef705729bd0637fc" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [ { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 0 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 1 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 2 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 3 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 4 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 5 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 6 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 7 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 8 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 9 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 10 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 11 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 12 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 13 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 14 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 15 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 16 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 17 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 18 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 19 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 20 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 21 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 22 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 23 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 24 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 25 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 26 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 27 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 28 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 29 }, { "source": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "source_row_zero_based": 30 } ], "artifact_integrity": {}, "resources": { "sample_count": 18, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 12804.0, "board_vram_min_free_mib": 11323.0, "host_ram_min_available_kib": 87180412.0, "process_rss_peak_kib": 80094944.0, "cgroup_peak_bytes_sampled": 9983438848.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C069" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/results.jsonl", "bytes": 1136, "sha256": "18bbe5a2addd0cf5b4069582a1bf7734b5912997241e2c76248ea3854459897a" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 95, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10676.0, "board_vram_min_free_mib": 13451.0, "host_ram_min_available_kib": 87925904.0, "process_rss_peak_kib": 78678880.0, "cgroup_peak_bytes_sampled": 851369984.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C070" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/results.jsonl", "bytes": 1137, "sha256": "d987ae554284f51ae337a8240dc330b95614803937a9c4e2dfe042800b352d0e" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 62, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10738.0, "board_vram_min_free_mib": 13389.0, "host_ram_min_available_kib": 87911700.0, "process_rss_peak_kib": 78674556.0, "cgroup_peak_bytes_sampled": 855879680.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C071" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/results.jsonl", "bytes": 1137, "sha256": "0acc280b50cfbfbb574c9418f456eca8ffdb35be4929a62da218abc9811fd49c" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 41, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10860.0, "board_vram_min_free_mib": 13267.0, "host_ram_min_available_kib": 87900112.0, "process_rss_peak_kib": 78699380.0, "cgroup_peak_bytes_sampled": 867377152.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C072" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/results.jsonl", "bytes": 1136, "sha256": "e65b653539096b7ab7b75b8a250ef9e323fbc7d74b98f9c48370c7efa9fea252" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 28, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 11108.0, "board_vram_min_free_mib": 13019.0, "host_ram_min_available_kib": 87899868.0, "process_rss_peak_kib": 78737760.0, "cgroup_peak_bytes_sampled": 893526016.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C073" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/results.jsonl", "bytes": 1135, "sha256": "497cc81965d54ad11a2b30385c5556b0e5ce3c4ef2e5c49c95ba8d3d9f11ecbe" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 95, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10676.0, "board_vram_min_free_mib": 13451.0, "host_ram_min_available_kib": 87930740.0, "process_rss_peak_kib": 78679396.0, "cgroup_peak_bytes_sampled": 853209088.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C074" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/results.jsonl", "bytes": 1136, "sha256": "c5c782ca8b6a7f64c8e6035acbc47dfccc1cb1c699487dc94aada1a0fa1655ea" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 63, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10738.0, "board_vram_min_free_mib": 13389.0, "host_ram_min_available_kib": 87937012.0, "process_rss_peak_kib": 78672700.0, "cgroup_peak_bytes_sampled": 857501696.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C075" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/results.jsonl", "bytes": 1137, "sha256": "4173fb748d85397af9a100546c4ebf763b52de7bd280ca311421c02b81661e68" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 41, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10860.0, "board_vram_min_free_mib": 13267.0, "host_ram_min_available_kib": 87906092.0, "process_rss_peak_kib": 78700204.0, "cgroup_peak_bytes_sampled": 870367232.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C076" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/results.jsonl", "bytes": 1135, "sha256": "24fa719439a54421148231b6365df83782ade33fa185293f0c3923d33062a191" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 104, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10676.0, "board_vram_min_free_mib": 13451.0, "host_ram_min_available_kib": 87925368.0, "process_rss_peak_kib": 78648852.0, "cgroup_peak_bytes_sampled": 2375077888.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C077" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/results.jsonl", "bytes": 1136, "sha256": "460a1232b5ca8ae8d0ccd0a61650911df8f8ab69ae7c1db75378682896b93786" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 63, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10738.0, "board_vram_min_free_mib": 13389.0, "host_ram_min_available_kib": 87919852.0, "process_rss_peak_kib": 78673624.0, "cgroup_peak_bytes_sampled": 853737472.0 } }, { "id": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C078" ], "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/results.jsonl", "bytes": 1136, "sha256": "5877447e02a26a5bf68db4e07da2b070635dfa0f4be4b64383c04e40fee43c7a" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 41, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10860.0, "board_vram_min_free_mib": 13267.0, "host_ram_min_available_kib": 87878360.0, "process_rss_peak_kib": 78697672.0, "cgroup_peak_bytes_sampled": 863432704.0 } }, { "id": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "summary": {}, "configuration_ids": [ "C079" ], "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 160, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 10760.0, "board_vram_min_free_mib": 13367.0, "host_ram_min_available_kib": 87840356.0, "process_rss_peak_kib": 78771044.0, "cgroup_peak_bytes_sampled": 8604721152.0 } }, { "id": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_PP8192_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C080" ], "source": { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/results.jsonl", "bytes": 1141, "sha256": "79b73c0a2b608355a47aa15a708ffe165066e9808e5aee0e332306c3dc2443bd" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 60, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21338.0, "board_vram_min_free_mib": 2789.0, "host_ram_min_available_kib": 86804680.0, "process_rss_peak_kib": 80909636.0, "cgroup_peak_bytes_sampled": 2576203776.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C081" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/results.jsonl", "bytes": 2274, "sha256": "2b18017dda26cf600bd5a52c15ddbc61b72ccfe2e26fbb20e7f0a6cbea27556d" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 87, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 14360.0, "board_vram_min_free_mib": 9767.0, "host_ram_min_available_kib": 87238284.0, "process_rss_peak_kib": 78923884.0, "cgroup_peak_bytes_sampled": 2443616256.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C082" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/results.jsonl", "bytes": 2270, "sha256": "75e37c24a1918c44ec3fa4f380427beca7fb1ea8013785f0046ecd977eb807c2" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 70, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 14360.0, "board_vram_min_free_mib": 9767.0, "host_ram_min_available_kib": 87256008.0, "process_rss_peak_kib": 78929168.0, "cgroup_peak_bytes_sampled": 2498899968.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C083" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/results.jsonl", "bytes": 2270, "sha256": "254b2c28d7aa3ee4cf4d89817705c962b8d71704c6f5cb5922dad1cca717b117" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 62, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16664.0, "board_vram_min_free_mib": 7463.0, "host_ram_min_available_kib": 87037916.0, "process_rss_peak_kib": 79060416.0, "cgroup_peak_bytes_sampled": 2662891520.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C084" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/results.jsonl", "bytes": 2269, "sha256": "76dfec7805cae6e5518ce01791891a0a570db1e154fff436831d51c016651368" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 70, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16654.0, "board_vram_min_free_mib": 7473.0, "host_ram_min_available_kib": 87253988.0, "process_rss_peak_kib": 78991168.0, "cgroup_peak_bytes_sampled": 2501140480.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C085" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/results.jsonl", "bytes": 2270, "sha256": "2025e9cd8dfff3b553dda790ef88aede231b4b2f30e6248540f9b7572f8c14f9" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 70, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 16654.0, "board_vram_min_free_mib": 7473.0, "host_ram_min_available_kib": 87214664.0, "process_rss_peak_kib": 78990872.0, "cgroup_peak_bytes_sampled": 2502959104.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C086" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/results.jsonl", "bytes": 2268, "sha256": "830767544026a37f5e44b2b28c2a22a7b4d23df73fa35a8f7bb76cd4eecc9bd7" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 61, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18958.0, "board_vram_min_free_mib": 5169.0, "host_ram_min_available_kib": 87099664.0, "process_rss_peak_kib": 79116040.0, "cgroup_peak_bytes_sampled": 2666311680.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C087" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/results.jsonl", "bytes": 2270, "sha256": "db1d56c38bdd6a7dde6dc9df216375fd06a402f9c926d86ab2c3a858ece4a902" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 69, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18950.0, "board_vram_min_free_mib": 5177.0, "host_ram_min_available_kib": 87209864.0, "process_rss_peak_kib": 79042992.0, "cgroup_peak_bytes_sampled": 2505228288.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C088" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/results.jsonl", "bytes": 2270, "sha256": "ceac905722f6af1cd4d5b1858a2f7cdc8137aed25ef81e1e5d0d31620555d10e" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 69, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 18950.0, "board_vram_min_free_mib": 5177.0, "host_ram_min_available_kib": 87235564.0, "process_rss_peak_kib": 79042816.0, "cgroup_peak_bytes_sampled": 2506190848.0 } }, { "id": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "native-bench", "status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "summary": {}, "configuration_ids": [ "C089" ], "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/results.jsonl", "bytes": 2269, "sha256": "c5c590ce00a30246e20ae85cf3b26d124faaebc74bdad7f22ac207a14aec41d7" }, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 61, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 21254.0, "board_vram_min_free_mib": 2873.0, "host_ram_min_available_kib": 87032556.0, "process_rss_peak_kib": 79210656.0, "cgroup_peak_bytes_sampled": 2670821376.0 } }, { "id": "profile-iq3-175-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "profiling-artifacts", "status": "PASS_STOCK_IQ3_PROFILE_NOT_DEPLOYED", "summary": { "status": "PASS_STOCK_IQ3_PROFILE_NOT_DEPLOYED", "started_utc": "2026-10-05T10:59:43.359389+00:00", "ended_utc": "2026-10-05T11:01:54.225214+00:00", "wall_seconds": 130.86585206608288, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716" }, "configuration_ids": [ "C090", "C091", "C092" ], "source": { "path": "profile-iq3-175-r1-20261005/summary.json", "bytes": 9010, "sha256": "c336d60b426acb3ce73b9d26a841b945eb1851595ff147c73f1defb139d5e66f" }, "summary_source": { "path": "profile-iq3-175-r1-20261005/summary.json", "bytes": 9010, "sha256": "c336d60b426acb3ce73b9d26a841b945eb1851595ff147c73f1defb139d5e66f" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 132, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 13048.0, "board_vram_min_free_mib": 11079.0, "host_ram_min_available_kib": 86287740.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "historical180/profile-iq3-175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "profiling-artifacts", "status": "FAIL_NO_RETRY", "summary": { "status": "FAIL_NO_RETRY", "started_utc": "2026-10-05T10:57:31.572986+00:00", "ended_utc": "2026-10-05T10:58:15.789948+00:00", "wall_seconds": 44.21698721998837, "failure": { "category": "QAError", "message": "Native trace failed; no inference retry" }, "engine_commit": "27c54b4bbcefadedcec6397477cc2e866c1db716" }, "configuration_ids": [ "C093" ], "source": { "path": "historical180/profile-iq3-175-20261005/summary.json", "bytes": 8227, "sha256": "d1ed74111150099b799821e7ebcfdf5dd3a820f91ad47752023d3e4b56b47c5f" }, "summary_source": { "path": "historical180/profile-iq3-175-20261005/summary.json", "bytes": 8227, "sha256": "d1ed74111150099b799821e7ebcfdf5dd3a820f91ad47752023d3e4b56b47c5f" }, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": { "integrity": [ { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "bytes": 10946624, "sha256": "268f81fdedf3149a538f252308927a4d5d1f6e062c178568a51e3b519744f8a8" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00002-of-00003.gguf", "bytes": 49567921344, "sha256": "cfe600b236b88c7fad1613a5ca5e83b9f2beb63cbd44c32b2be50a44747c695f" }, { "path": "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00003-of-00003.gguf", "bytes": 32382955968, "sha256": "f1912ba34c79427d2295a58dcb2b732b5931af5bef7a373c60557a57d9ee7250" }, { "path": "/mmproj-F16.gguf", "bytes": 904004000, "sha256": "1f7b7f0b984cf065c604360c29c8098362ed61b290db0ff12c6f360bb1a8a980" }, { "path": "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "bytes": 2786568256, "sha256": "5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6" }, { "path": "/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "/llama-server", "bytes": 17920, "sha256": "09132e554cb67ccc4ab9f75ed68c7974b10f8f0fe5c0b5d6494453cfcdc836d8" }, { "path": "/libcublas.so.12", "bytes": 116388640, "sha256": "031ce6c2cbfbb9468f040527cab5c599069ce5609e73e28f87503881063eac21" }, { "path": "/libcublasLt.so.12", "bytes": 751771728, "sha256": "10b5e6631cf8115c661eb895ed1533826308b58f7956466f53d236a40c9b622c" }, { "path": "/libcudart.so.12", "bytes": 728800, "sha256": "c3a75b33af334a3486d197dbd1584a2985183ba4688d237a2be5f2f679329920" }, { "path": "/libggml-base.so.0.22.0", "bytes": 981184, "sha256": "84755465eacb323d7ff06b1b7d3464429fd0c8690bafebbeb402481a4cd6a6fd" }, { "path": "/libggml-cpu.so.0.22.0", "bytes": 1425944, "sha256": "682c7e5a35465e35b7597f563a288076ed2b028bfb7d333b1cad2d0659c6f9a1" }, { "path": "/libggml-cuda.so.0.22.0", "bytes": 72214552, "sha256": "b9bd691276b9fd5ea02ffe25ab0fd63d17753c18dc6066c85a53fa3a3b271616" }, { "path": "/libggml.so.0.22.0", "bytes": 56016, "sha256": "e1bd31ff4386f56fa7606b16e073c56d531e90a1d822c9f2d376b23becd5a31c" }, { "path": "/libllama-bench-impl.so", "bytes": 489568, "sha256": "d70b6c888ab57ff1dda979457f480be2fbf6d6368f997aa96046a2ac0eaebc90" }, { "path": "/libllama-common.so.0.3.0", "bytes": 5954352, "sha256": "877abe3beaf38c84b9509b45a085dccb37d020bed682b3920e1d092124dc90c8" }, { "path": "/libllama-server-impl.so", "bytes": 3865680, "sha256": "33680d3d91af6df556de79f3dd759ad181bea965bf2edfe2fa1f5a438e066387" }, { "path": "/libllama.so.0.3.0", "bytes": 4760112, "sha256": "6a58e3a80d48d3af7297e192b6cbf63c583c6ba064ff178ad07b6cc65d63930a" }, { "path": "/libmtmd.so.0.3.0", "bytes": 1960872, "sha256": "7ebeda6be8676a1e555d123dc2bbad34535f3271f284537983a84592180dd5c8" }, { "path": "/libnccl.so.2", "bytes": 383583952, "sha256": "42ba6acaf7c1211480668f252c71801c8f54c72424f59f8a4d5462f394796ad6" }, { "path": "/llama-bench", "bytes": 17920, "sha256": "f9917cf9cff48a427b12e55d9c0c2972365df8989d84aa98c6f15e0055e5486c" }, { "path": "/llama-moe-trace", "bytes": 40472, "sha256": "b985ca6c65cbf13a73b833334b13336dd47a31ff7c1e2cd522a760d8d79bdf8e" } ] }, "resources": { "sample_count": 46, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 57.0, "board_vram_min_free_mib": 24070.0, "host_ram_min_available_kib": 88272092.0, "process_rss_peak_kib": null, "cgroup_peak_bytes_sampled": null } }, { "id": "historical180/profiles-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "type": "profiling-artifacts", "status": "NO_SUMMARY_STATUS", "summary": {}, "configuration_ids": [], "source": null, "summary_source": null, "missing_configuration": false, "invalid_nonmeasurement_lines": [], "artifact_integrity": {}, "resources": { "sample_count": 83, "scope": "run-wide sampled extrema; GPU board and host available RAM, not per-request process allocation", "board_vram_peak_mib": 9368.0, "board_vram_min_free_mib": 14759.0, "host_ram_min_available_kib": 87672468.0, "process_rss_peak_kib": 75802168.0, "cgroup_peak_bytes_sampled": 81967407104.0 } }, { "id": "strata188", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "type": "server-requests", "status": "MAX_CONTEXT_THROUGHPUT_MEASURED_PRODUCTION_RESTORED", "source": { "path": "jarvis-flash174/report/strata188/receipts/results-r3/benchmark/results.json", "bytes": 10102, "sha256": "4f51786ec9a1902e0dfc20283735e541d6ca2999f4b2bef5ac781645b868c5b4" }, "configuration_ids": [ "C094" ], "summary": { "status": "MAX_CONTEXT_THROUGHPUT_MEASURED_PRODUCTION_RESTORED", "strata_tag": "v0.1.40.1", "engine_version": "0.1.40", "physical_context": 262144, "actual_input": 260000, "input_fill_percent": 99.18212890625, "speed": { "prefill_tok_s": { "median": 1230.0384860118604, "min": 1224.690706717193, "max": 1247.3936668863998 }, "decode_tok_s": { "median": 30.397482708463208, "min": 29.292963967365807, "max": 30.547467901293494 }, "client_ttft_seconds": { "median": 211.67140187695622, "min": 208.74525400507264, "max": 212.61209778301418 } }, "resources": { "owner_vram_cap_MiB": 12288, "sampled_process_peak_MiB": 12168, "sampled_process_peak_GiB": 11.8828125, "sampled_board_used_peak_MiB": 12364.0, "host_MemAvailable_min_GiB": 10.544246673583984, "unit_MemoryPeak_bytes_live": 79521480704, "unit_swap_peak_bytes_live": 0, "measurement": "NVML sampled every ~2s; not a driver-enforced GPU quota. Unit peaks from live cgroup snapshots.", "host_io_caveat": "diskstats/page-fault/swap counters are host-wide; native file_MB is expert reads, not necessarily NVMe reads" }, "limitations": [ "Not same-quantization comparison with production Unsloth UD-IQ3_XXS.", "Synthetic fresh-input throughput test, not natural retrieval or an agent/voice acceptance.", "Thinking HIGH, greedy, capped at512 tokens; length finish may contain only reasoning, not a final answer.", "Q4 stock pack rounds195 projections to BF16; target pack is experimentally supported.", "No new compatible GSQ Q3 download, no application/engine source patch, no API promotion.", "Preparatory failures preserved: no Linux prebuilt, unsupported host CUDA installer, container bridge DNS, early target-state read, pre-existing empty log check." ], "production_restore": "ORIGINAL_Q3_262144_AUDIO_READY_ADMISSION_OPEN" }, "resources": { "sampled_process_peak_MiB": 12168, "sampled_process_peak_GiB": 11.8828125, "sampled_board_used_peak_MiB": 12364.0, "host_MemAvailable_min_GiB": 10.544246673583984, "unit_MemoryPeak_bytes_live": 79521480704, "unit_swap_peak_bytes_live": 0, "measurement": "NVML sampled every ~2s; not a driver-enforced GPU quota. Unit peaks from live cgroup snapshots.", "host_io_caveat": "diskstats/page-fault/swap counters are host-wide; native file_MB is expert reads, not necessarily NVMe reads" } }, { "id": "strata188-ram48", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "type": "server-requests", "status": "OWNER_INTERRUPTED", "source": { "path": "jarvis-flash174/report/strata188-ram48/receipts/results-ram48-r2/benchmark/results.json", "bytes": 4324, "sha256": "7e17aab4c46498826c125c80df08681b3b2d1b265c4339f50b201d3584581af1" }, "configuration_ids": [ "C095" ], "summary": { "status": "OWNER_INTERRUPTED", "strata_tag": "v0.1.40.1", "engine_version": "0.1.40", "physical_context": 262144, "actual_input": 260000, "input_fill_percent": 99.18212890625, "completed_maximum_context_runs": 1, "speed": { "prefill_tok_s": { "median": 773.1300437442926, "min": 773.1300437442926, "max": 773.1300437442926 }, "decode_tok_s": { "median": 9.300399263233995, "min": 9.300399263233995, "max": 9.300399263233995 }, "client_ttft_seconds": { "median": 336.625337178004, "min": 336.625337178004, "max": 336.625337178004 } }, "resources": { "memory_max_GiB": 48, "resident_budget_GiB": 24, "memory_peak_bytes": 42386587648, "memory_peak_GiB": 39.4755859375, "swap_peak_bytes": 0, "vram_limit_MiB": 12288, "sampled_process_peak_MiB": 12168, "sampled_process_peak_GiB": 11.8828125, "last_cgroup": { "memory.current": "7256727552", "memory.peak": "42386587648", "memory.max": "51539607552", "memory.swap.current": "0", "memory.swap.peak": "0", "memory.swap.max": "0", "memory.oom.group": "1", "memory.events": { "low": "0", "high": "0", "max": "0", "oom": "0", "oom_kill": "0", "oom_group_kill": "0", "sock_throttled": "0" }, "memory.stat": { "anon": "213463040", "file": "7041884160", "kernel": "880640", "kernel_stack": "32768", "pagetables": "618496", "sec_pagetables": "0", "percpu": "544", "sock": "0", "vmalloc": "0", "shmem": "0", "zswap": "0", "zswapped": "0", "file_mapped": "544768", "file_dirty": "4096", "file_writeback": "0", "swapcached": "0", "anon_thp": "0", "file_thp": "0", "shmem_thp": "0", "inactive_anon": "0", "active_anon": "213458944", "inactive_file": "6790230016", "active_file": "251658240", "unevictable": "0", "slab_reclaimable": "45368", "slab_unreclaimable": "177096", "slab": "222464", "hugetlb": "0", "workingset_refault_anon": "0", "workingset_refault_file": "0", "workingset_activate_anon": "0", "workingset_activate_file": "0", "workingset_restore_anon": "0", "workingset_restore_file": "0", "workingset_nodereclaim": "0", "pgdemote_kswapd": "0", "pgdemote_direct": "0", "pgdemote_khugepaged": "0", "pgdemote_proactive": "0", "pgpromote_success": "0", "pgscan": "0", "pgsteal": "0", "pswpin": "0", "pswpout": "0", "pgscan_kswapd": "0", "pgscan_direct": "0", "pgscan_khugepaged": "0", "pgscan_proactive": "0", "pgsteal_kswapd": "0", "pgsteal_direct": "0", "pgsteal_khugepaged": "0", "pgsteal_proactive": "0", "pgfault": "10887506", "pgmajfault": "5221", "pgrefill": "0", "pgactivate": "0", "pgdeactivate": "0", "pglazyfree": "0", "pglazyfreed": "0", "swpin_zero": "0", "swpout_zero": "0", "zswpin": "0", "zswpout": "0", "zswpwb": "0", "thp_fault_alloc": "0", "thp_collapse_alloc": "0", "thp_swpout": "0", "thp_swpout_fallback": "0", "numa_pages_migrated": "0", "numa_pte_updates": "0", "numa_hint_faults": "0" } }, "runtime_last_properties": { "ActiveState": "inactive", "Result": "success", "MemoryCurrent": "[not set]", "MemoryPeak": "[not set]", "MemorySwapCurrent": "[not set]", "MemorySwapPeak": "[not set]", "MemoryMax": "infinity", "MemorySwapMax": "infinity" } }, "limitations": [ "Linux cgroup48GiB plus charged file cache, not a physical48GiB host emulator.", "Native12GiB VRAM reserve and ~2s NVML guard, not a driver-enforced GPU partition.", "Synthetic fresh-input throughput, HIGH thinking,512-token length cap; not agent/voice or natural answer acceptance.", "Existing Q4, not same-quantization comparison with production UD-IQ3_XXS.", "Preparation attempt rejected unknown MemoryOOMGroup before GPU; corrected to stock OOMPolicy=kill.", "Owner stopped trial2 after one full maximum-context run. This is not a complete three-run benchmark.", "Original operator refused restoration after the stopped transient unit was garbage-collected; a separate hash-checked operator restored unchanged Q3/audio.", "Native expert/PLE direct I/O recorded; host NVMe/swap/pagefault counters are host-wide.", "Unchanged stock source, pack and weights; no download or engine/app patch." ], "original_operator_status": "FAILED_OR_ABORTED", "original_restore_error": "CalledProcessError: Command '('systemctl', 'stop', 'jarvis-strata188-runtime-ram48-r2.service')' returned non-zero exit status 5.", "production_restore": "ORIGINAL_Q3_262144_AUDIO_READY_ADMISSION_OPEN" }, "resources": { "sampled_process_peak_MiB": 12168, "sampled_process_peak_GiB": 11.8828125, "memory_max_GiB": 48, "resident_budget_GiB": 24, "memory_peak_GiB": 39.4755859375, "swap_peak_bytes": 0 } }, { "id": "functional195", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "type": "functional-transport", "status": "PASS_TRANSPORT_NOT_BENCHMARK", "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "configuration_ids": [ "C096" ], "summary": { "date": "2026-10-07", "not_performance_benchmark": true, "synthetic_repetitive_input": true, "repetitions": 1, "quality_or_speed_target_accepted": false, "notes": [ "Native printed decode counters are preserved rather than recomputed using an inconsistent token denominator.", "No memory/page-fault/NVMe sampling during these functional tests; do not invent benchmark telemetry.", "Prefill and decode from a repetitive synthetic fixture or a reused 99-token suffix are not general workload benchmarks.", "Audio services were inadvertently stopped by the earlier member-maintenance transaction during these requests; restored later, so these are not joint voice throughput measurements.", "Not publication-reviewed. Historical performance series counts remain an immutable older snapshot." ] } } ], "configurations": [ { "id": "C001", "run": "context176-262144-cpu48-q8-20261005", "source": { "path": "context176-262144-cpu48-q8-20261005/ctx262144-cpu48-q8_0-b8192/configuration.json", "bytes": 2147, "sha256": "2d7c52b17b7a4b0a5182ea8b4189b12d74e36636a0e9f4a319195fa1fc300251" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C002", "run": "context176-262144-cpu48-q8-u1024-20261005", "source": { "path": "context176-262144-cpu48-q8-u1024-20261005/ctx262144-cpu48-q8_0-b8192-u1024/configuration.json", "bytes": 2147, "sha256": "e2cb8a7ff9f14b47988a588a4b56e9f69541b1fec30f6ef8bc37a24265b03139" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "1024", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C003", "run": "context176-262144-cpu48-q8-u2048-20261005", "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/configuration.json", "bytes": 2147, "sha256": "eb0739664396eaa29e7e1196ccfe94a8e203ab42d109add7f99a0280d6c4b4f6" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "2048", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C004", "run": "context176-32768-cpu43-q8-20261005", "source": { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/configuration.json", "bytes": 2146, "sha256": "bfbc368e38b5ef2bb4a85a4d66565236e05d7dedcfc41fce73c27f1a03e0d771" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C005", "run": "context176-32768-cpu48-q8-control-20261005", "source": { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/configuration.json", "bytes": 2146, "sha256": "5c7c5471de304f68ca306df546ec5590a1ed9ef63bd3e493a50d35ab8219d045" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C006", "run": "context176-32768-cpu48-q8-u1024-control-20261005", "source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/configuration.json", "bytes": 2146, "sha256": "43566c28a2aad347a8e5366fe4772e14c128dbff6ee99de7a6eff188162e357c" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "1024", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C007", "run": "context176-32768-cpu48-q8-u2048-control-20261005", "source": { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/configuration.json", "bytes": 2146, "sha256": "480938ce5744ef74cc07f20bae6df84fa65c0a178e94f25ae882b4ee12f7c611" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "2048", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C008", "run": "context176-40960-cpu44-f16-u8192-screen-20261005", "source": { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "491e906764b52c775796c720a618cbe18d53c4b0c9522f7732eee01d1a956a24" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "44", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "40960", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C009", "run": "context176-45056-cpu45-f16-u8192-screen-20261005", "source": { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "1b5c0f89909e03f461d8901faa280f36c78c6c5d5bae7d880e652ea3a99259a3" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "45", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "45056", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C010", "run": "context176-47104-cpu45-f16-u8192-screen-20261005", "source": { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "9b8f25602bf689fd026f00bffb2ce94dd0e09c61987c42833db4af0b7ff02e68" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "45", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "47104", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C011", "run": "context176-49152-cpu46-f16-u8192-screen-20261005", "source": { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "ffee4c421befbc2f9bb01336a5a00e386a02ef6cba5be48f523ab29641bab8bc" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "46", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "49152", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C012", "run": "context176-65536-cpu48-f16-u8192-screen-20261005", "source": { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "0c02056115b63d74a5cbb37a2f75254dda81f84e2d59424a8c8e002a6bed2def" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14027", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "48", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "65536", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C013", "run": "guided-speed175-20261005", "source": { "path": "guided-speed175-20261005/baseline96/configuration.json", "bytes": 1622, "sha256": "f5f2a2fdad1dfa2182df479de95d4d672536266e6b3577edace51c083646f840" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "96", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "4096", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C014", "run": "guided-speed175-20261005", "source": { "path": "guided-speed175-20261005/batch8192-cache96/configuration.json", "bytes": 1622, "sha256": "087814832b2ab620091c561a4a98892c9bb19f07e6ad8c34d8a0167e031f4cd4" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "96", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C015", "run": "guided-speed175-20261005", "source": { "path": "guided-speed175-20261005/gpu-mtp2-cache96/configuration.json", "bytes": 2118, "sha256": "a485cd73d5fda457a43f623ec7551a264d41533b6c75a30ea5094713144b20d4" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "96", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "99", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "4096", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C016", "run": "guided-speed175-20261005", "source": { "path": "guided-speed175-20261005/overlap-cache112/configuration.json", "bytes": 1595, "sha256": "741ea1c27bf242763b06503282eaf126c7d27035a18d3c65e991f6145ab75b84" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "112", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "4096", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C017", "run": "guided-speed175-finalist-20261005", "source": { "path": "guided-speed175-finalist-20261005/batch8192-cache80/configuration.json", "bytes": 1650, "sha256": "fb9c4186871f22d7bce37aad37f53dcfa9fbf0c4bda8f66d124bee4c38e959db" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C018", "run": "guided-speed175-followup-20261005", "source": { "path": "guided-speed175-followup-20261005/batch8192-cache80/configuration.json", "bytes": 1622, "sha256": "6b8f16237300e7125def1b702d0931a9956af3fde08bcb92f027eb4c27765a71" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C019", "run": "guided-speed175-followup-20261005", "source": { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/configuration.json", "bytes": 2118, "sha256": "f737871efe0e520a93facd3131f74a9e684b9f0a970904afb9ec7b8bf343a661" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "64", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "99", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "4096", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C020", "run": "guided-speed175-followup-20261005", "source": { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/configuration.json", "bytes": 2117, "sha256": "59e4fe17c871bf55f6b827a75c62acbb3e7b68e5c06bc39d8c82d88822679adc" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "72", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "99", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "0", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C021", "run": "guided-speed175-native-profile-20261005", "source": { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/configuration.json", "bytes": 1652, "sha256": "1db4a961601d87d10a75d3b22ff824e977d6c7ba9224914d67bb18e2f174fb5d" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "16", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "16", "--threads-batch", "16", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "16384", "--ubatch-size", "16384", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C022", "run": "guided-speed175-native-profile-20261005", "source": { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/configuration.json", "bytes": 1650, "sha256": "fb9c4186871f22d7bce37aad37f53dcfa9fbf0c4bda8f66d124bee4c38e959db" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C023", "run": "guided-speed175-native-profile-20261005", "source": { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/configuration.json", "bytes": 1650, "sha256": "6e1cdb8f05539023e9ae6f1adf3918a4f9dc7dca5470ad551fa13ac88b716853" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "16", "--threads-batch", "16", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C024", "run": "historical180/mtp-screen32-20261005", "source": { "path": "historical180/mtp-screen32-20261005/control/configuration.json", "bytes": 1795, "sha256": "b15bb1ecb94ff5b10bd8fad553f478c17de1775ef37f7ef94af1cff4c04fb150" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14024", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "99", "--fit", "off", "--spec-type", "none", "--load-mode", "mmap", "--lazy-mode", "on", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--threads", "14", "--threads-batch", "14", "--batch-size", "4096", "--ubatch-size", "4096", "--no-sched-async-cpu", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--cache-prompt", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "64", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--offline" ], "environment": { "LANG": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C025", "run": "http-baseline6-cache32-vision-20261004", "source": { "path": "http-baseline6-cache32-vision-20261004/configuration.json", "bytes": 1424, "sha256": "8c8829a8ee0b31c41962533e19d9b0a63b15fdba5e9d67fe24a53c1eb3236f03" }, "argv": [], "environment": {} }, { "id": "C026", "run": "http-batch1024-cache32-vision-20261005", "source": { "path": "http-batch1024-cache32-vision-20261005/configuration.json", "bytes": 3691, "sha256": "3d371913a816640ec1877fc7cbdc45bc7460396dbb7a37f4be456c5518e7fd2b" }, "argv": [], "environment": {} }, { "id": "C027", "run": "http-context32-cache32-u1024-20261005", "source": { "path": "http-context32-cache32-u1024-20261005/configuration.json", "bytes": 3628, "sha256": "53b018526e9b800739dce80320aab9d1f4873f549440c1609ff274e79c6756cd" }, "argv": [], "environment": {} }, { "id": "C028", "run": "http-threads14-cache32-vision-20261004", "source": { "path": "http-threads14-cache32-vision-20261004/configuration.json", "bytes": 1389, "sha256": "f2a316eff1d06613abb33796f06a6809a7fe92b1fe3a51c45acee4758447681e" }, "argv": [], "environment": {} }, { "id": "C029", "run": "ik178/results/control32-noMTP", "source": { "path": "ik178/results/control32-noMTP/summary.json", "bytes": 5586, "sha256": "9b9be58d9282469e9c2aba8149c325b6a58226fe8d4ed47a845edb502d3be76c" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14029", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-mmap", "--defer-ple", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1", "--top-p", ".95", "--top-k", "20", "--min-p", "0", "--presence-penalty", "0", "--repeat-penalty", "1", "--jinja", "--metrics", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C030", "run": "ik178/results/mtp32-n1", "source": { "path": "ik178/results/mtp32-n1/summary.json", "bytes": 5169, "sha256": "3a806194e6fb5f40482c99875e9d4515b995d654195f487ffff2c8e3c82fedba" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14029", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-mmap", "--defer-ple", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1", "--top-p", ".95", "--top-k", "20", "--min-p", "0", "--presence-penalty", "0", "--repeat-penalty", "1", "--jinja", "--metrics", "--no-context-shift", "--timeout", "7200", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-type", "mtp:n_max=1,p_min=0.0", "--cache-type-k-draft", "f16", "--cache-type-v-draft", "f16", "--threads-draft", "14", "--threads-batch-draft", "14" ], "environment": {} }, { "id": "C031", "run": "ik178/results/mtp32-n1-draftU512", "source": { "path": "ik178/results/mtp32-n1-draftU512/summary.json", "bytes": 5590, "sha256": "6136b95fc0c866ba4e61fbf3487405d6ee451b17ea050a65a16c10a5ddcc7bcc" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14029", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-mmap", "--defer-ple", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1", "--top-p", ".95", "--top-k", "20", "--min-p", "0", "--presence-penalty", "0", "--repeat-penalty", "1", "--jinja", "--metrics", "--no-context-shift", "--timeout", "7200", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-type", "mtp:n_max=1,p_min=0.0", "--cache-type-k-draft", "f16", "--cache-type-v-draft", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--draft-params", "--ubatch-size 512" ], "environment": {} }, { "id": "C032", "run": "ik178/results/mtp32-n1-mainU4096-draftU512", "source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/summary.json", "bytes": 6255, "sha256": "37c8d10eb66e50bd0774a19f93ab523e39249403d4c38acdafd0988ebff3986d" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14029", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-mmap", "--defer-ple", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--batch-size", "8192", "--ubatch-size", "4096", "--temp", "1", "--top-p", ".95", "--top-k", "20", "--min-p", "0", "--presence-penalty", "0", "--repeat-penalty", "1", "--jinja", "--metrics", "--no-context-shift", "--timeout", "7200", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-type", "mtp:n_max=1,p_min=0.0", "--cache-type-k-draft", "f16", "--cache-type-v-draft", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--draft-params", "--ubatch-size 512" ], "environment": {} }, { "id": "C033", "run": "iq3-author32-accept-20261005", "source": { "path": "iq3-author32-accept-20261005/author-control/configuration.json", "bytes": 1594, "sha256": "3e804083f04371e0fd259245f4c8e7327d54917604cdd3d8237aaa59d90d84f4" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14024", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "48", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "6", "--threads-batch", "6", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "2048", "--ubatch-size", "512", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C034", "run": "iq3-author32-accept-20261005", "source": { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/configuration.json", "bytes": 1919, "sha256": "c5173776d3d6a455bd55c3b9bdc541505bbd39a1e90a6da1832a17183ce1e487" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14024", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "48", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "0", "--spec-type", "draft-mtp", "--spec-draft-n-max", "1", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "6", "--threads-batch", "6", "--threads-draft", "6", "--threads-batch-draft", "6", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "2048", "--ubatch-size", "512", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C035", "run": "iq3-cache96-accept-20261005", "source": { "path": "iq3-cache96-accept-20261005/cache96-control/configuration.json", "bytes": 1594, "sha256": "9d078f937789dd7fcbacb68dcd9e5be6b4c9b1d57ed111bce03e2f8561eef3b8" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14024", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "96", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "6", "--threads-batch", "6", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "2048", "--ubatch-size", "512", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C036", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/configuration.json", "bytes": 1597, "sha256": "e388672909f4b19d48eabc736a2a03fee5bf24de9684c83e019b47db4a259d59" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14025", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "96", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "4096", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C037", "run": "opt181/results/opt256-cache4g-ple-overlap-workspace", "source": { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/summary.json", "bytes": 7148, "sha256": "d81e16005e2f64a28c0d74ed8db3ed02166a0840fd0895471a30b9ea7f3b9d4c" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14030", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "48", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200", "--moe-expert-cache-mib", "4096", "--phase-aware-workspace", "--live-context-workspace", "--ple-prefetch", "--backend-sampling", "--decode-overlap" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C038", "run": "placement-speed175-20261005", "source": { "path": "placement-speed175-20261005/placement-control-mmap/configuration.json", "bytes": 1650, "sha256": "fb9c4186871f22d7bce37aad37f53dcfa9fbf0c4bda8f66d124bee4c38e959db" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C039", "run": "placement-speed175-20261005", "source": { "path": "placement-speed175-20261005/placement-none-lazy/configuration.json", "bytes": 1650, "sha256": "a143594576931cc78494a307d838d0347f5521095448220c455e98ad540f42ce" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C040", "run": "placement-speed175-r1-20261005", "source": { "path": "placement-speed175-r1-20261005/placement-none-lazy/configuration.json", "bytes": 1650, "sha256": "a143594576931cc78494a307d838d0347f5521095448220c455e98ad540f42ce" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C041", "run": "placement-speed175-r2-20261005", "source": { "path": "placement-speed175-r2-20261005/placement-none-lazy/configuration.json", "bytes": 1650, "sha256": "a143594576931cc78494a307d838d0347f5521095448220c455e98ad540f42ce" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "80", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C042", "run": "prefill-dma175-20261005", "source": { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/configuration.json", "bytes": 2173, "sha256": "ac4d4df09305f0d1ea50f9cdfd7559869fd93c5ccc0d9c450d82fabec915b75a" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "42", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C043", "run": "prefill-finalist175-20261005", "source": { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/configuration.json", "bytes": 2171, "sha256": "efcf44f8d07f15a7be2b26bf68ad5d3bf578976ac2138fa6887b72c012f4aded" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C044", "run": "prefill-finalist175-20261005", "source": { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/configuration.json", "bytes": 2173, "sha256": "12424a7e77d263f241d71201b3b444bc4a48910c534f9a56b66db1836b1941b2" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C045", "run": "prefill-hybrid175-20261005", "source": { "path": "prefill-hybrid175-20261005/uncached-hybrid38/configuration.json", "bytes": 1649, "sha256": "c48fbe8bfcb4ad25e7d1d3be3c5917c398cd08658abce552421eec11f4a8bef7" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "38", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C046", "run": "prefill-hybrid175-20261005", "source": { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/configuration.json", "bytes": 2173, "sha256": "9328fdac61cc49145f4c01e2ce8a7027410b353089688b2013c6798d440eee8f" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "42", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "2", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C047", "run": "prefill-kv175-20261005", "source": { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/configuration.json", "bytes": 2171, "sha256": "59c0bdd22f64f5c7e277be3c4a09f4ad1ab33607d2671c5de72ed528b58a76a8" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "3", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C048", "run": "prefill-kv175-20261005", "source": { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/configuration.json", "bytes": 2173, "sha256": "be9f2454261d69f8c89d06cf18b9dd83fa40920bffedc1a3e52e5546c8ab2bb9" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "3", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C049", "run": "prefill-mtp1-finalist175-20261005", "source": { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/configuration.json", "bytes": 2171, "sha256": "6c6690ef429d3cfea4c6be1478e6cd4d845910b0011fa79555987b5cae772fd8" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "draft-mtp", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--model-draft", "/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf", "--gpu-layers-draft", "99", "--spec-draft-n-max", "1", "--spec-draft-n-min", "0", "--spec-draft-type-k", "f16", "--spec-draft-type-v", "f16", "--threads-draft", "14", "--threads-batch-draft", "14", "--cpu-mask-draft", "0xffff", "--cpu-strict-draft", "1", "--cpu-mask-batch-draft", "0xffff", "--cpu-strict-batch-draft", "1" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C050", "run": "prefill-path175-20261005", "source": { "path": "prefill-path175-20261005/uncached-cpu99/configuration.json", "bytes": 1649, "sha256": "1f8498aaa5c872a416120bd1dbd5ba24f7b701126233813fa09b7deb5642f35c" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14026", "--moe-cache-profile", "/routing-merged.csv", "--moe-cache-slots", "0", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "99", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "mmap", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0" } }, { "id": "C051", "run": "upstream178/results/control32-noMTP", "source": { "path": "upstream178/results/control32-noMTP/summary.json", "bytes": 6496, "sha256": "5c0debeab732d308cd55543ee0d9dd135d76dde4915bbcd1eedf0b36f47cb23f" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--moe-cache-profile", "/routing-iq3-175.csv", "--moe-cache-slots", "0", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "43", "--no-sched-async-cpu", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C052", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/summary.json", "bytes": 6930, "sha256": "46a8cfacb8e8799cc45e0c8a11a29d64ebef40ef180086b84f77ea781d07df2c" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "43", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "2048", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C053", "run": "upstream178/results/upstream256-cpu46-u4096-deep179", "source": { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/summary.json", "bytes": 6373, "sha256": "364bfb1390edb821e2b87e1921af4f340f17d0c7f291d447b793c03437bbdf27" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "46", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C054", "run": "upstream178/results/upstream256-cpu46-u4096-short179", "source": { "path": "upstream178/results/upstream256-cpu46-u4096-short179/summary.json", "bytes": 6367, "sha256": "9b0adac2f2dbff4be04fddef2aa90b094caafd2f1c217a6711956c6453044ef0" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "46", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C055", "run": "upstream178/results/upstream256-f16-noMTP", "source": { "path": "upstream178/results/upstream256-f16-noMTP/summary.json", "bytes": 6932, "sha256": "c999cba80d5ade26cfb4b9e93f63295854dd7118644150a8cc7967b10c881c6a" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "48", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C056", "run": "upstream178/results/upstream256-noMTP", "source": { "path": "upstream178/results/upstream256-noMTP/summary.json", "bytes": 6933, "sha256": "d6f234b71ffcd674d84abb60a8e9585f00628fe063ee50b2d449d661108e431b" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "48", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "4096", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C057", "run": "upstream178/results/upstream32-noMTP", "source": { "path": "upstream178/results/upstream32-noMTP/summary.json", "bytes": 6642, "sha256": "2bdd58a9329b42a8d1800349601c64370732f230d6809f4622dd91d5a33bf7bc" }, "argv": [ "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14028", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "43", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "32768", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "8192", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200" ], "environment": {} }, { "id": "C058", "run": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888", "source": { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/provenance.txt", "bytes": 2123, "sha256": "a75b75c63cacb66eb5eaf2fc0b3f1a52323ea4337b17bc0fdd67e264ed453b56" }, "provenance_parameters": "utc=2026-10-04T21:43:18Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=cache\nslots=32\nthreads=14\nasync_cpu=1\nmin_gpu_free_mib=4656\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 512 --sched-async-cpu 1 -p 512 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C059", "run": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674", "source": { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/provenance.txt", "bytes": 2123, "sha256": "1cc229adab87896c90605bffdbe26a22ad5920446af4cf3ad546adf81eb7b42d" }, "provenance_parameters": "utc=2026-10-04T21:45:01Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=cache\nslots=40\nthreads=14\nasync_cpu=1\nmin_gpu_free_mib=4656\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=40\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 512 --sched-async-cpu 1 -p 512 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C060", "run": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724", "source": { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/provenance.txt", "bytes": 2123, "sha256": "bdea10d216c872f6e161212eb146c8e1f87be49194ad90b82403af34971b8e4f" }, "provenance_parameters": "utc=2026-10-04T21:46:53Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=cache\nslots=48\nthreads=14\nasync_cpu=1\nmin_gpu_free_mib=4656\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 512 --sched-async-cpu 1 -p 512 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C061", "run": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507", "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/provenance.txt", "bytes": 2125, "sha256": "4028072afa973d2cd48dc5db011c0da2873ca854b641766c9e1b076f948d207a" }, "provenance_parameters": "utc=2026-10-04T21:48:42Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=overlap\nslots=48\nthreads=14\nasync_cpu=0\nmin_gpu_free_mib=4656\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 512 --sched-async-cpu 0 -p 512 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C062", "run": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180", "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/provenance.txt", "bytes": 2125, "sha256": "b298e7d289fcf5f9e5f0d43a6ba7565a740f2664e942d0506ad0401057f1178d" }, "provenance_parameters": "utc=2026-10-04T21:50:31Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=overlap\nslots=48\nthreads=14\nasync_cpu=1\nmin_gpu_free_mib=4656\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 512 --sched-async-cpu 1 -p 512 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C063", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/provenance.txt", "bytes": 2560, "sha256": "10eb52c1c889a53ebbf2c780c20a4db623d6ed09709bc78336616aba484d9b28" }, "provenance_parameters": "utc=2026-10-04T22:37:34Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 1024 -ub 1024 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C064", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/provenance.txt", "bytes": 2558, "sha256": "60ddd1ff29bbc3f859c6ba9386653f02813d30c39bc8e3285ea7c065504f8b56" }, "provenance_parameters": "utc=2026-10-04T22:30:48Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 1024 -ub 128 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C065", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/provenance.txt", "bytes": 2558, "sha256": "5e50859a1dda06bbe235fb57b53cb95e66454ea021c70713db9fb1893a31b12c" }, "provenance_parameters": "utc=2026-10-04T22:34:03Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 1024 -ub 256 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C066", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/provenance.txt", "bytes": 2558, "sha256": "83ca1734bde8768fddb14ac16f62346b944830adc10b85ec83f20d664e4d33f8" }, "provenance_parameters": "utc=2026-10-04T22:36:10Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 1024 -ub 512 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C067", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/provenance.txt", "bytes": 2560, "sha256": "67954303ce91a62534d67d9bb34c85fac95a6d5fd2dc9344f79def7fdebfa1f0" }, "provenance_parameters": "utc=2026-10-04T22:45:15Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 1024 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C068", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/provenance.txt", "bytes": 2526, "sha256": "ded5755269689995cbfaa4c720f855164ecbb34a966534f372343147d9e0a11c" }, "provenance_parameters": "utc=2026-10-05T06:06:34Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 1024 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C069", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/provenance.txt", "bytes": 2558, "sha256": "9c5e0797536c61dc31b2b3e52b0753750642f463c5230af34ed4b0d12f5ef6e6" }, "provenance_parameters": "utc=2026-10-04T22:38:30Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 128 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C070", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/provenance.txt", "bytes": 2558, "sha256": "7fafec9133078a09d3b4c43fef6821fd41ab8a0f76f21b2b0c4abd7c41fa7c0d" }, "provenance_parameters": "utc=2026-10-04T22:41:44Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 256 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C071", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/provenance.txt", "bytes": 2558, "sha256": "7b31b74c73c4830d3dd4308ecc2f828c8d2529acd0e6802c4a366481c46d975d" }, "provenance_parameters": "utc=2026-10-04T22:43:51Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 512 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C072", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/provenance.txt", "bytes": 2560, "sha256": "50d2c7c3d472cc46ed00e3708febe6648a4248ab4779cdfd9d00ab92b3a24d94" }, "provenance_parameters": "utc=2026-10-04T22:52:56Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 1024 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C073", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/provenance.txt", "bytes": 2558, "sha256": "2fdbfe20ee2e6c7a6f89ff4c3775842acef90689bf9d5ce6337a2f0d363c5269" }, "provenance_parameters": "utc=2026-10-04T22:46:11Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 128 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C074", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/provenance.txt", "bytes": 2558, "sha256": "3da7a5d4f4d8fd0a785e9dc784390077fed1874976d86337143957b86826eb91" }, "provenance_parameters": "utc=2026-10-04T22:49:25Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 256 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C075", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/provenance.txt", "bytes": 2558, "sha256": "227772b1dfa903c8a503f7e0a76118d382a6c169ab9ae1fbffaacaa61647a4d2" }, "provenance_parameters": "utc=2026-10-04T22:51:32Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 512 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C076", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/provenance.txt", "bytes": 2556, "sha256": "97a389fef86bcf01e82875ade5af70877cf084acd9be0808aebb2ff2571324fa" }, "provenance_parameters": "utc=2026-10-04T22:23:19Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 512 -ub 128 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C077", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/provenance.txt", "bytes": 2556, "sha256": "d584ba4aaee934d05890bb082e13fde9c0ebd17f96d982bf668389199d3b489b" }, "provenance_parameters": "utc=2026-10-04T22:27:16Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 512 -ub 256 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C078", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700", "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/provenance.txt", "bytes": 2556, "sha256": "52eae6326dfa74e90830f7061d34417f161256147c90e25b6a45592815e57185" }, "provenance_parameters": "utc=2026-10-04T22:29:24Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=prefill\nslots=32\nthreads=14\nasync_cpu=1\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=9058\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nproduction_context_and_projector_extra_reserve_mib=4402\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_CUDA_ENABLE_UNIFIED_MEMORY\nunset=GGML_SCHED_PREFETCH_EXPERTS\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 512 -ub 512 --sched-async-cpu 1 -p 4096 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C079", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/provenance.txt", "bytes": 2154, "sha256": "a1a07a5796fcd6e50e178220c8f210b548dd0f9ed531e5ecebe2058bf9132219" }, "provenance_parameters": "utc=2026-10-04T21:36:54Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=threads\nslots=32\nthreads=4,6,8,12,14,16\nasync_cpu=1\nmin_gpu_free_mib=4656\nmin_mem_available_kib=8388608\nstock_bench_short_context=true\nproduction_context_not_measured=131072\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=2\nresource_extrema_are_sampled=true\ninput=stock_random_tokens\nMTP=not_requested\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=32\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 4\\,6\\,8\\,12\\,14\\,16 -b 2048 -ub 512 --sched-async-cpu 1 -p 512 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C080", "run": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850", "source": { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/provenance.txt", "bytes": 2652, "sha256": "76b7cff805c169c65d80bb8158651957579d595466c4c3c40a4057600527e837" }, "provenance_parameters": "utc=2026-10-05T07:39:38Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=pp8192\nslots=48\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_physical_context=8192\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 8192 -ub 8192 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 8192 -n 0 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C081", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/provenance.txt", "bytes": 2669, "sha256": "8de5eb652b9a96fb7ab771036360e232b4fcdd490c8cafcf74b011412a9d7ae1" }, "provenance_parameters": "utc=2026-10-05T06:44:46Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=48\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 2048 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C082", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/provenance.txt", "bytes": 2669, "sha256": "329446d3530a905f50e044c5d5d05ad36d470b71c5f1ed53270619fff03c894d" }, "provenance_parameters": "utc=2026-10-05T06:46:17Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=48\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 2048 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C083", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/provenance.txt", "bytes": 2669, "sha256": "152d2cc76bdebe4ca487e23e9298de41c3e1c325b32c611b0e5113b6df0c996a" }, "provenance_parameters": "utc=2026-10-05T06:47:30Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=48\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=48\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 4096 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C084", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/provenance.txt", "bytes": 2669, "sha256": "cd1b90794e92d64eac23a19147306fe9ba74ac99d97821ac64fed28538571086" }, "provenance_parameters": "utc=2026-10-05T06:48:35Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=64\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=64\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 2048 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C085", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/provenance.txt", "bytes": 2669, "sha256": "f0377fe7569f5ddd616ccaeb0d7fb78697d80a754dadb7b680d06d5d35fa447d" }, "provenance_parameters": "utc=2026-10-05T06:49:48Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=64\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=64\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 2048 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C086", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/provenance.txt", "bytes": 2669, "sha256": "7deecda7d7122c7e6ce7c260d8e9d4654d136dec0cf9cf797cc19cf220c75693" }, "provenance_parameters": "utc=2026-10-05T06:51:01Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=64\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=64\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 4096 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C087", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/provenance.txt", "bytes": 2669, "sha256": "10bc97868340c30498eddc10904bd017aa0176f4a8a51fcf679088bf43e2637c" }, "provenance_parameters": "utc=2026-10-05T06:52:05Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=80\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=80\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 2048 -ub 2048 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C088", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/provenance.txt", "bytes": 2669, "sha256": "11f777426880217c84a2b9e54efa850654c935af256d2d3b3f57845309cb57c8" }, "provenance_parameters": "utc=2026-10-05T06:53:18Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=80\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=80\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 2048 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C089", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277", "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/provenance.txt", "bytes": 2669, "sha256": "a0bbd27aba47bb47a2a617483fede87239d30d32cd2a6ddc477d10cfe45494da" }, "provenance_parameters": "utc=2026-10-05T06:54:30Z\nsource_commit=27c54b4bbcefadedcec6397477cc2e866c1db716\nphase=speed\nslots=80\nthreads=14\nasync_cpu=0\nstock_bench_threads_batch_supported=false\nstock_bench_t_sets_both_thread_counts=true\nmin_gpu_free_mib=2048\nmin_mem_available_kib=8388608\nMTP=not_requested\nstock_bench_short_contexts=4096_and_256_separate\nrepetitions=3\nwarmup=stock_default\nOS_page_cache_not_dropped=true\nresource_sampling_seconds=1\nresource_extrema_are_sampled=true\ninput=stock_random_token_ids\nGGML_MOE_CACHE_PROFILE=/routing-merged.csv\nGGML_MOE_CACHE_SLOTS=80\nLD_LIBRARY_PATH=/bin\nunset=GGML_CUDA_REGISTER_HOST,GGML_SCHED_PREFETCH_EXPERTS,GGML_CUDA_ENABLE_UNIFIED_MEMORY,GGML_OP_OFFLOAD_MIN_BATCH\ncommand=/llama-bench -m /Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf --offline -ngl 99 -ncmoe 99 -lm mmap -lzm on -fa on -ctk q8_0 -ctv q8_0 -C 0xffff --cpu-strict 1 -t 14 -b 4096 -ub 4096 --sched-async-cpu 0 --no-op-offload 0 -d 0 -p 4096 -n 256 -r 3 -o jsonl --progress --verbose ", "argv": [], "environment": {} }, { "id": "C090", "run": "profile-iq3-175-r1-20261005", "source": { "path": "profile-iq3-175-r1-20261005/chat.configuration.json", "bytes": 1314, "sha256": "0bfcdd54605c7874477e266e07d26c668d51fb060870b67f0bc3a2ba33bd4e11" }, "argv": [ "/llama-moe-trace", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--offline", "--fit", "off", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "99", "--load-mode", "mmap", "--lazy-mode", "on", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--batch-size", "8192", "--ubatch-size", "8192", "--moe-cache-slots", "0", "--no-escape", "--file", "/chat.prompt.txt", "--n-predict", "512" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0", "MOE_TRACE_OUT": "/chat.csv" } }, { "id": "C091", "run": "profile-iq3-175-r1-20261005", "source": { "path": "profile-iq3-175-r1-20261005/code.configuration.json", "bytes": 1314, "sha256": "dfcb2a4679ff720432d2cea8d88d70bf5119be1737066b787f022cb101bb16b8" }, "argv": [ "/llama-moe-trace", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--offline", "--fit", "off", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "99", "--load-mode", "mmap", "--lazy-mode", "on", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--batch-size", "8192", "--ubatch-size", "8192", "--moe-cache-slots", "0", "--no-escape", "--file", "/code.prompt.txt", "--n-predict", "512" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0", "MOE_TRACE_OUT": "/code.csv" } }, { "id": "C092", "run": "profile-iq3-175-r1-20261005", "source": { "path": "profile-iq3-175-r1-20261005/tools.configuration.json", "bytes": 1316, "sha256": "e2f3a79012392d2102af27dd3a6627a755d8774ac43962bac5df000716bbea73" }, "argv": [ "/llama-moe-trace", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--offline", "--fit", "off", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "99", "--load-mode", "mmap", "--lazy-mode", "on", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--batch-size", "8192", "--ubatch-size", "8192", "--moe-cache-slots", "0", "--no-escape", "--file", "/tools.prompt.txt", "--n-predict", "512" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0", "MOE_TRACE_OUT": "/tools.csv" } }, { "id": "C093", "run": "historical180/profile-iq3-175-20261005", "source": { "path": "historical180/profile-iq3-175-20261005/code.configuration.json", "bytes": 1339, "sha256": "cd00480ee4e11c8c6b73511fde9fb2ae43c67c9036d23a2733ef1042118cbe8a" }, "argv": [ "/llama-moe-trace", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--offline", "--fit", "off", "--ctx-size", "32768", "--parallel", "1", "--gpu-layers", "99", "--n-cpu-moe", "99", "--load-mode", "mmap", "--lazy-mode", "on", "--flash-attn", "on", "--cache-type-k", "q8_0", "--cache-type-v", "q8_0", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--batch-size", "8192", "--ubatch-size", "8192", "--moe-cache-slots", "0", "--spec-type", "none", "--no-escape", "--file", "/code.prompt.txt", "--n-predict", "512" ], "environment": { "LANG": "pl_PL.UTF-8", "LC_ALL": "C.UTF-8", "LD_LIBRARY_PATH": "/bin", "CUDA_DEVICE_ORDER": "PCI_BUS_ID", "CUDA_VISIBLE_DEVICES": "0", "MOE_TRACE_OUT": "/code.csv" } }, { "id": "C094", "run": "strata188", "source": { "path": "jarvis-flash174/report/strata188/summary.json", "bytes": 7164, "sha256": "208db26d4d20937ae638a1dee16b498ce9902aef909e834224ed87402ef69781" }, "argv": [ "/strata", "--pack", "/unsloth-ud-q4_k_xl", "--native", "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf", "--expert-profile", "/expert-profile.bin", "--expert-cache", "auto", "--prefill", "auto", "--spec", "4", "--spec-min-p", "0.5", "--mtp", "/rt", "--max-context", "262144", "--kv", "int8", "--kv-resident", "32768", "--resident-budget-gib", "62", "--vram-reserve-mib", "12288" ], "environment": {}, "runtime_effective": { "workers": 15, "host_threads": 1, "cpu_affinity": "host0; pool1-15", "gpu_expert_slots": 1407, "gpu_expert_cache_GiB": 4.09, "prompt_chunk_tokens": 7424, "prompt_borrowed_slots": 1185, "kv_pinned_host_GiB": 3.09, "kv_gpu_resident_cells_per_layer": 32768, "resident_expert_budget_GiB": 62, "mtp_expert_quantization": "q2_0", "spec_draft_max": 4, "spec_min_p": 0.5, "draft_vocabulary_tokens": 106299 }, "systemd_MemoryMax_GiB": 80, "systemd_MemorySwapMax_GiB": 0 }, { "id": "C095", "run": "strata188-ram48", "source": { "path": "jarvis-flash174/report/strata188-ram48/summary.json", "bytes": 8999, "sha256": "99909693256ef583e4ff73f4ac20b68e2a6c7981e25e4baf6d836e720d90e0e7" }, "argv": [ "/strata", "--pack", "/unsloth-ud-q4_k_xl", "--native", "/Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf", "--expert-profile", "/expert-profile.bin", "--expert-cache", "auto", "--prefill", "auto", "--spec", "4", "--spec-min-p", "0.5", "--mtp", "/rt", "--max-context", "262144", "--kv", "int8", "--kv-resident", "32768", "--resident-budget-gib", "24", "--vram-reserve-mib", "12288" ], "environment": {}, "runtime_effective": {}, "systemd_MemoryMax_GiB": 48, "systemd_MemorySwapMax_GiB": 0 }, { "id": "C096", "run": "functional195", "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "argv": [ "/usr/bin/env", "LD_LIBRARY_PATH=/bin", "/llama-server", "--model", "/Qwen3.8-Flash-Next-UD-IQ3_XXS-00001-of-00003.gguf", "--alias", "jarvis-flash-next", "--host", "127.0.0.1", "--port", "14004", "--spec-type", "none", "--gpu-layers", "99", "--n-cpu-moe", "48", "--threads", "14", "--threads-batch", "14", "--cpu-mask", "0xffff", "--cpu-strict", "1", "--load-mode", "none", "--lazy-mode", "on", "--fit", "off", "--flash-attn", "on", "--cache-type-k", "f16", "--cache-type-v", "f16", "--ctx-size", "262144", "--parallel", "1", "--cache-reuse", "256", "--cache-prompt", "--batch-size", "8192", "--ubatch-size", "2048", "--temp", "1.0", "--top-p", "0.95", "--top-k", "20", "--min-p", "0.0", "--presence-penalty", "0.0", "--repeat-penalty", "1.0", "--metrics", "--slots", "--jinja", "--chat-template-file", "/chat_template.jinja", "--chat-template-kwargs", "{\"reasoning_effort\":\"xhigh\"}", "--verbosity", "4", "--offline", "--no-context-shift", "--timeout", "7200", "--mmproj", "/mmproj-F16.gguf", "--no-mmproj-offload" ], "environment": {}, "vendor_template_commit": "855bffc49448e299789730ff92c9b8d834d6cc14", "vendor_template_sha256": "e57684bae4156211a55473c5a63be976a405a37ab5be5ae0e5abf1df5349c4b2" } ], "measurements": [ { "id": "M0001", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 665.8102379052465, "decode_tps": 31.796648795000348, "first_frame_s": 12.306223395979032, "first_answer_s": null, "wall_s": 28.377586888964288, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:11:24.925193+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 12303.806, "prompt_per_token_ms": 1.501929443359375, "prompt_per_second": 665.8102379052465, "predicted_n": 512, "predicted_ms": 16070.876, "predicted_per_token_ms": 31.449855185909982, "predicted_per_second": 31.796648795000348, "draft_n": 430, "draft_n_accepted": 296 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6883720930232559, "mean_accepted_span_including_target_bonus": 2.376744186046512, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7581395348837209, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6186046511627907 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 430.0, "llamacpp:spec_decode_num_accepted_tokens_total": 296.0, "llamacpp:spec_decode_num_drafts_total": 215.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 163.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 133.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0002", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 65536, "output_tokens": 512, "fresh_input_tokens": 65536, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 608.1173515052932, "decode_tps": 27.07132451673447, "first_frame_s": 107.78033383807633, "first_answer_s": null, "wall_s": 126.65814240800682, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:13:31.745451+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 65536, "prompt_ms": 107768.673, "prompt_per_token_ms": 1.644419448852539, "prompt_per_second": 608.1173515052932, "predicted_n": 512, "predicted_ms": 18876.062, "predicted_per_token_ms": 36.93945596868885, "predicted_per_second": 27.07132451673447, "draft_n": 421, "draft_n_accepted": 300 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 65536, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7125890736342043, "mean_accepted_span_including_target_bonus": 2.4218009478672986, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7962085308056872, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6255924170616114 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 421.0, "llamacpp:spec_decode_num_accepted_tokens_total": 300.0, "llamacpp:spec_decode_num_drafts_total": 211.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 132.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "769470ad08861dadfc117b6f0f8660d91319dab585ca5d6102b6a632c6b20442", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0003", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 65536, "output_tokens": 233, "fresh_input_tokens": 65536, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 606.5993428649044, "decode_tps": 28.494384948086793, "first_frame_s": 108.72050832107197, "first_answer_s": 115.33334618411027, "wall_s": 116.86412126210053, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:15:28.768011+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 65536, "prompt_ms": 108038.363, "prompt_per_token_ms": 1.6485345916748047, "prompt_per_second": 606.5993428649044, "predicted_n": 233, "predicted_ms": 8141.955, "predicted_per_token_ms": 35.094633620689656, "predicted_per_second": 28.494384948086793, "draft_n": 180, "draft_n_accepted": 143 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 65536, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7944444444444444, "mean_accepted_span_including_target_bonus": 2.5888888888888886, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8444444444444444, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7444444444444445 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 180.0, "llamacpp:spec_decode_num_accepted_tokens_total": 143.0, "llamacpp:spec_decode_num_drafts_total": 90.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 76.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 67.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "4ac7642e4ce91fa9b013e393eee5a330f71dc848ad026e96e3619413424ef29f", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0004", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 512, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 509.26059376905414, "decode_tps": 21.998484386220547, "first_frame_s": 258.27812321006786, "first_answer_s": null, "wall_s": 281.5098050989909, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:20:10.564772+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 257377.071, "prompt_per_token_ms": 1.963631217956543, "prompt_per_second": 509.26059376905414, "predicted_n": 512, "predicted_ms": 23228.873, "predicted_per_token_ms": 45.4576771037182, "predicted_per_second": 21.998484386220547, "draft_n": 427, "draft_n_accepted": 297 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6955503512880562, "mean_accepted_span_including_target_bonus": 2.3878504672897196, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7663551401869159, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6214953271028038 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 427.0, "llamacpp:spec_decode_num_accepted_tokens_total": 297.0, "llamacpp:spec_decode_num_drafts_total": 214.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 133.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "f80d502c12228c3f16a2ec6a752fccd3523f5ffd3403b839d7767397ddfeb5db", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0005", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 251, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 508.0853492524313, "decode_tps": 23.168461311125466, "first_frame_s": 259.44006253394764, "first_answer_s": 268.3980342509458, "wall_s": 270.23348898696713, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:24:41.080750+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 257972.406, "prompt_per_token_ms": 1.9681732635498046, "prompt_per_second": 508.0853492524313, "predicted_n": 251, "predicted_ms": 10790.531, "predicted_per_token_ms": 43.162124000000006, "predicted_per_second": 23.168461311125466, "draft_n": 198, "draft_n_accepted": 153 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7727272727272727, "mean_accepted_span_including_target_bonus": 2.5454545454545454, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8181818181818182, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7272727272727273 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_total": 153.0, "llamacpp:spec_decode_num_drafts_total": 99.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 81.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 72.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "5cd57289870a0f5f9cb10260b7d1f7e5e502736c47eb279841c6245eab081bca", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0006", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 196608, "output_tokens": 512, "fresh_input_tokens": 196608, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 434.19881566425056, "decode_tps": 18.087611363140088, "first_frame_s": 454.442154959077, "first_answer_s": null, "wall_s": 482.6978467099834, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:32:44.181915+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 0, "prompt_n": 196608, "prompt_ms": 452806.394, "prompt_per_token_ms": 2.3030924173990885, "prompt_per_second": 434.19881566425056, "predicted_n": 512, "predicted_ms": 28251.381, "predicted_per_token_ms": 55.28645988258317, "predicted_per_second": 18.087611363140088, "draft_n": 432, "draft_n_accepted": 294 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 196608, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6805555555555556, "mean_accepted_span_including_target_bonus": 2.361111111111111, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7407407407407407, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6203703703703703 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 432.0, "llamacpp:spec_decode_num_accepted_tokens_total": 294.0, "llamacpp:spec_decode_num_drafts_total": 216.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 160.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 134.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "f7a798b7796d7025ebc5f079858226f94688c51bf546fd5d1ad0f837b94c328e", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0007", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 196608, "output_tokens": 275, "fresh_input_tokens": 196608, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 433.62171812442796, "decode_tps": 19.96212441882169, "first_frame_s": 455.3671379170846, "first_answer_s": 466.8974386010086, "wall_s": 469.09788025706075, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:40:33.705404+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 196608, "prompt_ms": 453409.024, "prompt_per_token_ms": 2.306157552083333, "prompt_per_second": 433.62171812442796, "predicted_n": 275, "predicted_ms": 13725.994, "predicted_per_token_ms": 50.094868613138686, "predicted_per_second": 19.96212441882169, "draft_n": 214, "draft_n_accepted": 169 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 196608, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7897196261682243, "mean_accepted_span_including_target_bonus": 2.5794392523364484, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8317757009345794, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7476635514018691 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 214.0, "llamacpp:spec_decode_num_accepted_tokens_total": 169.0, "llamacpp:spec_decode_num_drafts_total": 107.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 89.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 80.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "409a0e5ffece86e629ee48137437093f0ce9fa2a177efd2a25a42c6ea7d80607", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0008", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 512, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 402.4505334418032, "decode_tps": 15.796551788948673, "first_frame_s": 588.2418287720066, "first_answer_s": null, "wall_s": 620.5955197099829, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:50:54.791051+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 586231.053, "prompt_per_token_ms": 2.4847774245641694, "prompt_per_second": 402.4505334418032, "predicted_n": 512, "predicted_ms": 32348.832, "predicted_per_token_ms": 63.30495499021526, "predicted_per_second": 15.796551788948673, "draft_n": 457, "draft_n_accepted": 282 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6170678336980306, "mean_accepted_span_including_target_bonus": 2.2314410480349345, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7074235807860262, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5240174672489083 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 457.0, "llamacpp:spec_decode_num_accepted_tokens_total": 282.0, "llamacpp:spec_decode_num_drafts_total": 229.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 162.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 120.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "6945518a0f22640c75ce148211a7c5e064402e7a69fb5fe526f95d3b5221fef1", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0009", "run": "context176-262144-cpu48-q8-u2048-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx262144-cpu48-q8_0-b8192-u2048", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 215, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 402.1074549047437, "decode_tps": 19.37806551488516, "first_frame_s": 589.091108931927, "first_answer_s": 597.7256108259317, "wall_s": 600.1395734309917, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:00:55.446630+00:00", "configuration_ids": [ "C003" ], "missing_configuration_match": false, "source": { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 586731.226, "prompt_per_token_ms": 2.4868974394839127, "prompt_per_second": 402.1074549047437, "predicted_n": 215, "predicted_ms": 11043.414, "predicted_per_token_ms": 51.60473831775701, "predicted_per_second": 19.37806551488516, "draft_n": 156, "draft_n_accepted": 138 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8846153846153846, "mean_accepted_span_including_target_bonus": 2.769230769230769, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9102564102564102, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8589743589743589 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 156.0, "llamacpp:spec_decode_num_accepted_tokens_total": 138.0, "llamacpp:spec_decode_num_drafts_total": 78.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 71.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 67.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "273a52049c92150ef0914ca9783c589ef5d3fed19543fae15071c8a478002f26", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0010", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1045.0923143615228, "decode_tps": 31.559866472737152, "first_frame_s": 7.841722868965007, "first_answer_s": null, "wall_s": 24.032643829938024, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:42:06.488784+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7838.542, "prompt_per_token_ms": 0.956853271484375, "prompt_per_second": 1045.0923143615228, "predicted_n": 512, "predicted_ms": 16191.45, "predicted_per_token_ms": 31.685812133072407, "predicted_per_second": 31.559866472737152, "draft_n": 453, "draft_n_accepted": 284 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6269315673289183, "mean_accepted_span_including_target_bonus": 2.251101321585903, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7048458149779736, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5462555066079295 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 453.0, "llamacpp:spec_decode_num_accepted_tokens_total": 284.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 160.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0011", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1201.9184919678921, "decode_tps": 33.08371208572605, "first_frame_s": 6.817871488048695, "first_answer_s": null, "wall_s": 22.26368492504116, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:42:28.794555+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6815.77, "prompt_per_token_ms": 0.832003173828125, "prompt_per_second": 1201.9184919678921, "predicted_n": 512, "predicted_ms": 15445.667, "predicted_per_token_ms": 30.226354207436398, "predicted_per_second": 33.08371208572605, "draft_n": 455, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.621978021978022, "mean_accepted_span_including_target_bonus": 2.2412280701754383, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6973684210526315, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.543859649122807 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 455.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 228.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0012", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1197.2780048250654, "decode_tps": 33.04676207372506, "first_frame_s": 6.844161374960095, "first_answer_s": null, "wall_s": 22.307234012987465, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:42:51.143421+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6842.187, "prompt_per_token_ms": 0.8352279052734375, "prompt_per_second": 1197.2780048250654, "predicted_n": 512, "predicted_ms": 15462.937, "predicted_per_token_ms": 30.260150684931507, "predicted_per_second": 33.04676207372506, "draft_n": 455, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.621978021978022, "mean_accepted_span_including_target_bonus": 2.2412280701754383, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6973684210526315, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.543859649122807 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 455.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 228.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0013", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 196, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1182.4168039441222, "decode_tps": 37.76776619932123, "first_frame_s": 7.076125109102577, "first_answer_s": 11.05291939608287, "wall_s": 12.239428151049651, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:43:03.442595+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6928.183, "prompt_per_token_ms": 0.8457254638671875, "prompt_per_second": 1182.4168039441222, "predicted_n": 196, "predicted_ms": 5163.133, "predicted_per_token_ms": 26.477605128205127, "predicted_per_second": 37.76776619932123, "draft_n": 148, "draft_n_accepted": 123 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.831081081081081, "mean_accepted_span_including_target_bonus": 2.6621621621621623, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8783783783783784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7837837837837838 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 148.0, "llamacpp:spec_decode_num_accepted_tokens_total": 123.0, "llamacpp:spec_decode_num_drafts_total": 74.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 58.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0014", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1151.6589469996488, "decode_tps": 33.970471350594146, "first_frame_s": 14.419570673955604, "first_answer_s": null, "wall_s": 29.462445099954493, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:43:32.981148+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14226.434, "prompt_per_token_ms": 0.8683126220703125, "prompt_per_second": 1151.6589469996488, "predicted_n": 512, "predicted_ms": 15042.476, "predicted_per_token_ms": 29.437330724070453, "predicted_per_second": 33.970471350594146, "draft_n": 415, "draft_n_accepted": 303 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7301204819277108, "mean_accepted_span_including_target_bonus": 2.456730769230769, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7932692307692307, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6634615384615384 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 415.0, "llamacpp:spec_decode_num_accepted_tokens_total": 303.0, "llamacpp:spec_decode_num_drafts_total": 208.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 138.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0015", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1158.2818602272491, "decode_tps": 34.88936725025904, "first_frame_s": 14.148886015987955, "first_answer_s": null, "wall_s": 28.795539106009528, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:44:01.818484+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14145.089, "prompt_per_token_ms": 0.8633477172851562, "prompt_per_second": 1158.2818602272491, "predicted_n": 512, "predicted_ms": 14646.296, "predicted_per_token_ms": 28.662027397260275, "predicted_per_second": 34.88936725025904, "draft_n": 415, "draft_n_accepted": 303 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7301204819277108, "mean_accepted_span_including_target_bonus": 2.456730769230769, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7932692307692307, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6634615384615384 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 415.0, "llamacpp:spec_decode_num_accepted_tokens_total": 303.0, "llamacpp:spec_decode_num_drafts_total": 208.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 138.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0016", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1157.2891913399394, "decode_tps": 35.324364769759796, "first_frame_s": 14.161308586015366, "first_answer_s": null, "wall_s": 28.627600335981697, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:44:30.487456+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14157.222, "prompt_per_token_ms": 0.8640882568359375, "prompt_per_second": 1157.2891913399394, "predicted_n": 512, "predicted_ms": 14465.936, "predicted_per_token_ms": 28.309072407045008, "predicted_per_second": 35.324364769759796, "draft_n": 415, "draft_n_accepted": 303 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7301204819277108, "mean_accepted_span_including_target_bonus": 2.456730769230769, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7932692307692307, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6634615384615384 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 415.0, "llamacpp:spec_decode_num_accepted_tokens_total": 303.0, "llamacpp:spec_decode_num_drafts_total": 208.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 138.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0017", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 204, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1150.0749542258375, "decode_tps": 40.11986154497436, "first_frame_s": 14.567582688992843, "first_answer_s": 18.423182263970375, "wall_s": 19.627816381980665, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:44:50.190569+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14246.028, "prompt_per_token_ms": 0.869508544921875, "prompt_per_second": 1150.0749542258375, "predicted_n": 204, "predicted_ms": 5059.838, "predicted_per_token_ms": 24.925310344827587, "predicted_per_second": 40.11986154497436, "draft_n": 142, "draft_n_accepted": 133 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9366197183098591, "mean_accepted_span_including_target_bonus": 2.873239436619718, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9577464788732394, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.9154929577464789 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 142.0, "llamacpp:spec_decode_num_accepted_tokens_total": 133.0, "llamacpp:spec_decode_num_drafts_total": 71.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 68.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 65.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0018", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 24576, "output_tokens": 512, "fresh_input_tokens": 24576, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1117.6862802508301, "decode_tps": 30.992513094336786, "first_frame_s": 22.32292925193906, "first_answer_s": null, "wall_s": 38.81118137796875, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:45:29.090182+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 24576, "prompt_ms": 21988.281, "prompt_per_token_ms": 0.8947054443359375, "prompt_per_second": 1117.6862802508301, "predicted_n": 512, "predicted_ms": 16487.853, "predicted_per_token_ms": 32.26585714285714, "predicted_per_second": 30.992513094336786, "draft_n": 436, "draft_n_accepted": 292 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 24576, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6697247706422018, "mean_accepted_span_including_target_bonus": 2.3394495412844036, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7385321100917431, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6009174311926605 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 436.0, "llamacpp:spec_decode_num_accepted_tokens_total": 292.0, "llamacpp:spec_decode_num_drafts_total": 218.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 161.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "f5732903d8aa0d1e6cafe8efc8634c8a409f0d9d7e1fa2e49b4d31b664ec144a", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0019", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 24576, "output_tokens": 512, "fresh_input_tokens": 24576, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1120.3978141665953, "decode_tps": 32.63141520212704, "first_frame_s": 21.939873179071583, "first_answer_s": null, "wall_s": 37.60014598502312, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:46:06.736260+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 0, "prompt_n": 24576, "prompt_ms": 21935.066, "prompt_per_token_ms": 0.8925401204427083, "prompt_per_second": 1120.3978141665953, "predicted_n": 512, "predicted_ms": 15659.756, "predicted_per_token_ms": 30.64531506849315, "predicted_per_second": 32.63141520212704, "draft_n": 434, "draft_n_accepted": 293 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 24576, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6751152073732719, "mean_accepted_span_including_target_bonus": 2.3502304147465436, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7419354838709677, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6082949308755761 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 434.0, "llamacpp:spec_decode_num_accepted_tokens_total": 293.0, "llamacpp:spec_decode_num_drafts_total": 217.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 161.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 132.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "f5732903d8aa0d1e6cafe8efc8634c8a409f0d9d7e1fa2e49b4d31b664ec144a", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0020", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 24576, "output_tokens": 512, "fresh_input_tokens": 24576, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1119.907272261142, "decode_tps": 32.5844686990788, "first_frame_s": 21.94963638007175, "first_answer_s": null, "wall_s": 37.63238289405126, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:46:44.413284+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 0, "prompt_n": 24576, "prompt_ms": 21944.674, "prompt_per_token_ms": 0.8929310709635416, "prompt_per_second": 1119.907272261142, "predicted_n": 512, "predicted_ms": 15682.318, "predicted_per_token_ms": 30.68946771037182, "predicted_per_second": 32.5844686990788, "draft_n": 434, "draft_n_accepted": 293 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 24576, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6751152073732719, "mean_accepted_span_including_target_bonus": 2.3502304147465436, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7419354838709677, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6082949308755761 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 434.0, "llamacpp:spec_decode_num_accepted_tokens_total": 293.0, "llamacpp:spec_decode_num_drafts_total": 217.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 161.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 132.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "f5732903d8aa0d1e6cafe8efc8634c8a409f0d9d7e1fa2e49b4d31b664ec144a", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0021", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 24576, "output_tokens": 224, "fresh_input_tokens": 24576, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1113.2895313819358, "decode_tps": 37.8525035611907, "first_frame_s": 22.506341775064357, "first_answer_s": 27.11626453406643, "wall_s": 28.39805891108699, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:47:12.901484+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 24576, "prompt_ms": 22075.12, "prompt_per_token_ms": 0.8982389322916666, "prompt_per_second": 1113.2895313819358, "predicted_n": 224, "predicted_ms": 5891.288, "predicted_per_token_ms": 26.41833183856502, "predicted_per_second": 37.8525035611907, "draft_n": 160, "draft_n_accepted": 144 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 24576, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9, "mean_accepted_span_including_target_bonus": 2.8, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9375, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8625 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 160.0, "llamacpp:spec_decode_num_accepted_tokens_total": 144.0, "llamacpp:spec_decode_num_drafts_total": 80.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 75.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 69.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "92395f776c8641a6c377c1e80ca62f34467617edce260b3f606058f9cb75e6e1", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0022", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 29491, "output_tokens": 512, "fresh_input_tokens": 29491, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1075.214834238594, "decode_tps": 30.882397112211823, "first_frame_s": 27.86152444407344, "first_answer_s": null, "wall_s": 44.4088488050038, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:47:57.405276+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 0, "prompt_n": 29491, "prompt_ms": 27428.007, "prompt_per_token_ms": 0.9300466922111831, "prompt_per_second": 1075.214834238594, "predicted_n": 512, "predicted_ms": 16546.643, "predicted_per_token_ms": 32.380906066536205, "predicted_per_second": 30.882397112211823, "draft_n": 439, "draft_n_accepted": 290 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 29491, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6605922551252847, "mean_accepted_span_including_target_bonus": 2.3181818181818183, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7227272727272728, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5954545454545455 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 439.0, "llamacpp:spec_decode_num_accepted_tokens_total": 290.0, "llamacpp:spec_decode_num_drafts_total": 220.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "5ac7a8010c896ec914f1a16128efaa37c6c90c25eebe45a11de3b65f2cea2b25", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0023", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 29491, "output_tokens": 512, "fresh_input_tokens": 29491, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1073.671773976928, "decode_tps": 31.83720682901233, "first_frame_s": 27.47405566298403, "first_answer_s": null, "wall_s": 43.525038070976734, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:48:40.977164+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 0, "prompt_n": 29491, "prompt_ms": 27467.426, "prompt_per_token_ms": 0.9313833372893425, "prompt_per_second": 1073.671773976928, "predicted_n": 512, "predicted_ms": 16050.403, "predicted_per_token_ms": 31.409790606653623, "predicted_per_second": 31.83720682901233, "draft_n": 441, "draft_n_accepted": 289 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 29491, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.655328798185941, "mean_accepted_span_including_target_bonus": 2.3076923076923075, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7149321266968326, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5927601809954751 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 441.0, "llamacpp:spec_decode_num_accepted_tokens_total": 289.0, "llamacpp:spec_decode_num_drafts_total": 221.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 158.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "5ac7a8010c896ec914f1a16128efaa37c6c90c25eebe45a11de3b65f2cea2b25", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0024", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 29491, "output_tokens": 512, "fresh_input_tokens": 29491, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1074.890423393692, "decode_tps": 31.55023272464229, "first_frame_s": 27.44205797603354, "first_answer_s": null, "wall_s": 43.63900004304014, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:49:24.659827+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 0, "prompt_n": 29491, "prompt_ms": 27436.285, "prompt_per_token_ms": 0.930327388016683, "prompt_per_second": 1074.890423393692, "predicted_n": 512, "predicted_ms": 16196.394, "predicted_per_token_ms": 31.695487279843444, "predicted_per_second": 31.55023272464229, "draft_n": 441, "draft_n_accepted": 289 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 29491, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.655328798185941, "mean_accepted_span_including_target_bonus": 2.3076923076923075, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7149321266968326, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5927601809954751 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 441.0, "llamacpp:spec_decode_num_accepted_tokens_total": 289.0, "llamacpp:spec_decode_num_drafts_total": 221.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 158.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "5ac7a8010c896ec914f1a16128efaa37c6c90c25eebe45a11de3b65f2cea2b25", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0025", "run": "context176-32768-cpu43-q8-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu43-q8_0-b8192", "context_tokens": 32768, "input_tokens": 29491, "output_tokens": 209, "fresh_input_tokens": 29491, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1070.546008957215, "decode_tps": 36.44386383451122, "first_frame_s": 28.014693784993142, "first_answer_s": 32.452141942922026, "wall_s": 33.722821633913554, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:49:58.478898+00:00", "configuration_ids": [ "C004" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 29491, "prompt_ms": 27547.625, "prompt_per_token_ms": 0.9341027771184429, "prompt_per_second": 1070.546008957215, "predicted_n": 209, "predicted_ms": 5707.408, "predicted_per_token_ms": 27.43946153846154, "predicted_per_second": 36.44386383451122, "draft_n": 154, "draft_n_accepted": 133 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 29491, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8636363636363636, "mean_accepted_span_including_target_bonus": 2.7272727272727275, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9090909090909091, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8181818181818182 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 154.0, "llamacpp:spec_decode_num_accepted_tokens_total": 133.0, "llamacpp:spec_decode_num_drafts_total": 77.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 70.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 63.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "03978cfbfe577682fbb4b0f707edd9e8ca5cca61eec1ff5d7a4a99dad56bfda6", "complete_native_template": true, "requested_output_budget": 3245, "truncated": false } }, { "id": "M0026", "run": "context176-32768-cpu48-q8-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1015.5498634735029, "decode_tps": 30.914182592108837, "first_frame_s": 8.069807706982829, "first_answer_s": null, "wall_s": 24.599090337986127, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:54:13.747800+00:00", "configuration_ids": [ "C005" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-control-20261005/requests.json", "bytes": 34850, "sha256": "ff498936a7091a684a8353a86fd943613ec9e70fa85f1e8ba5db3637f7977a56" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8066.566, "prompt_per_token_ms": 0.984688232421875, "prompt_per_second": 1015.5498634735029, "predicted_n": 512, "predicted_ms": 16529.63, "predicted_per_token_ms": 32.347612524461844, "predicted_per_second": 30.914182592108837, "draft_n": 423, "draft_n_accepted": 299 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.706855791962175, "mean_accepted_span_including_target_bonus": 2.410377358490566, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7924528301886793, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6179245283018868 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 423.0, "llamacpp:spec_decode_num_accepted_tokens_total": 299.0, "llamacpp:spec_decode_num_drafts_total": 212.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0027", "run": "context176-32768-cpu48-q8-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1168.4097489758944, "decode_tps": 32.777341530533796, "first_frame_s": 7.0134642630582675, "first_answer_s": null, "wall_s": 22.603690642048605, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:54:36.392820+00:00", "configuration_ids": [ "C005" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-control-20261005/requests.json", "bytes": 34850, "sha256": "ff498936a7091a684a8353a86fd943613ec9e70fa85f1e8ba5db3637f7977a56" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7011.239, "prompt_per_token_ms": 0.8558641357421874, "prompt_per_second": 1168.4097489758944, "predicted_n": 512, "predicted_ms": 15590.038, "predicted_per_token_ms": 30.508880626223092, "predicted_per_second": 32.777341530533796, "draft_n": 423, "draft_n_accepted": 299 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.706855791962175, "mean_accepted_span_including_target_bonus": 2.410377358490566, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7924528301886793, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6179245283018868 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 423.0, "llamacpp:spec_decode_num_accepted_tokens_total": 299.0, "llamacpp:spec_decode_num_drafts_total": 212.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0028", "run": "context176-32768-cpu48-q8-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1164.9298660174231, "decode_tps": 32.45337478897368, "first_frame_s": 7.034104069927707, "first_answer_s": null, "wall_s": 22.780017565004528, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:54:59.216666+00:00", "configuration_ids": [ "C005" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-control-20261005/requests.json", "bytes": 34850, "sha256": "ff498936a7091a684a8353a86fd943613ec9e70fa85f1e8ba5db3637f7977a56" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7032.183, "prompt_per_token_ms": 0.8584207763671875, "prompt_per_second": 1164.9298660174231, "predicted_n": 512, "predicted_ms": 15745.666, "predicted_per_token_ms": 30.81343639921722, "predicted_per_second": 32.45337478897368, "draft_n": 423, "draft_n_accepted": 299 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.706855791962175, "mean_accepted_span_including_target_bonus": 2.410377358490566, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7924528301886793, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6179245283018868 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 423.0, "llamacpp:spec_decode_num_accepted_tokens_total": 299.0, "llamacpp:spec_decode_num_drafts_total": 212.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 131.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0029", "run": "context176-32768-cpu48-q8-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu48-q8_0-b8192", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 197, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1150.7310133948067, "decode_tps": 35.539824549315135, "first_frame_s": 7.276218491024338, "first_answer_s": 11.538166697020642, "wall_s": 12.79134758899454, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:55:12.066511+00:00", "configuration_ids": [ "C005" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-control-20261005/requests.json", "bytes": 34850, "sha256": "ff498936a7091a684a8353a86fd943613ec9e70fa85f1e8ba5db3637f7977a56" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7118.953, "prompt_per_token_ms": 0.8690128173828126, "prompt_per_second": 1150.7310133948067, "predicted_n": 197, "predicted_ms": 5514.94, "predicted_per_token_ms": 28.137448979591834, "predicted_per_second": 35.539824549315135, "draft_n": 148, "draft_n_accepted": 123 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.831081081081081, "mean_accepted_span_including_target_bonus": 2.6621621621621623, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8783783783783784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7837837837837838 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 148.0, "llamacpp:spec_decode_num_accepted_tokens_total": 123.0, "llamacpp:spec_decode_num_drafts_total": 74.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 58.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0030", "run": "context176-32768-cpu48-q8-u1024-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192-u1024", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 466.3847179755897, "decode_tps": 30.212648563752175, "first_frame_s": 17.567930140998214, "first_answer_s": null, "wall_s": 34.481106690946035, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:02:56.115305+00:00", "configuration_ids": [ "C006" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.json", "bytes": 42460, "sha256": "d0f8d9927daefdc84aee21bf558e7c953254a9a823ba1fc860270028e4019d02" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 17564.898, "prompt_per_token_ms": 2.144152587890625, "prompt_per_second": 466.3847179755897, "predicted_n": 512, "predicted_ms": 16913.446, "predicted_per_token_ms": 33.09872015655577, "predicted_per_second": 30.212648563752175, "draft_n": 432, "draft_n_accepted": 295 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6828703703703703, "mean_accepted_span_including_target_bonus": 2.3657407407407405, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.75, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6157407407407407 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 432.0, "llamacpp:spec_decode_num_accepted_tokens_total": 295.0, "llamacpp:spec_decode_num_drafts_total": 216.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 162.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 133.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0031", "run": "context176-32768-cpu48-q8-u1024-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192-u1024", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 498.8835106393668, "decode_tps": 32.94883143197298, "first_frame_s": 16.42313017998822, "first_answer_s": null, "wall_s": 31.93234585598111, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:03:28.090014+00:00", "configuration_ids": [ "C006" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.json", "bytes": 42460, "sha256": "d0f8d9927daefdc84aee21bf558e7c953254a9a823ba1fc860270028e4019d02" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 16420.667, "prompt_per_token_ms": 2.0044759521484377, "prompt_per_second": 498.8835106393668, "predicted_n": 512, "predicted_ms": 15508.896, "predicted_per_token_ms": 30.35009001956947, "predicted_per_second": 32.94883143197298, "draft_n": 430, "draft_n_accepted": 296 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6883720930232559, "mean_accepted_span_including_target_bonus": 2.376744186046512, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7534883720930232, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6232558139534884 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 430.0, "llamacpp:spec_decode_num_accepted_tokens_total": 296.0, "llamacpp:spec_decode_num_drafts_total": 215.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 162.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 134.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0032", "run": "context176-32768-cpu48-q8-u1024-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192-u1024", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 499.2529488955613, "decode_tps": 32.986829604821004, "first_frame_s": 16.410544923041016, "first_answer_s": null, "wall_s": 31.901839283062145, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:04:00.037447+00:00", "configuration_ids": [ "C006" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.json", "bytes": 42460, "sha256": "d0f8d9927daefdc84aee21bf558e7c953254a9a823ba1fc860270028e4019d02" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 16408.516, "prompt_per_token_ms": 2.00299267578125, "prompt_per_second": 499.2529488955613, "predicted_n": 512, "predicted_ms": 15491.031, "predicted_per_token_ms": 30.315129158512722, "predicted_per_second": 32.986829604821004, "draft_n": 430, "draft_n_accepted": 296 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6883720930232559, "mean_accepted_span_including_target_bonus": 2.376744186046512, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7534883720930232, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6232558139534884 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 430.0, "llamacpp:spec_decode_num_accepted_tokens_total": 296.0, "llamacpp:spec_decode_num_drafts_total": 215.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 162.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 134.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0033", "run": "context176-32768-cpu48-q8-u1024-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu48-q8_0-b8192-u1024", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 217, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 490.15583748557714, "decode_tps": 38.281629797853505, "first_frame_s": 16.915342204039916, "first_answer_s": 21.343371169059537, "wall_s": 22.558024447993375, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:04:22.654125+00:00", "configuration_ids": [ "C006" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.json", "bytes": 42460, "sha256": "d0f8d9927daefdc84aee21bf558e7c953254a9a823ba1fc860270028e4019d02" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 16713.052, "prompt_per_token_ms": 2.04016748046875, "prompt_per_second": 490.15583748557714, "predicted_n": 217, "predicted_ms": 5642.393, "predicted_per_token_ms": 26.122189814814814, "predicted_per_second": 38.281629797853505, "draft_n": 154, "draft_n_accepted": 140 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9090909090909091, "mean_accepted_span_including_target_bonus": 2.8181818181818183, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.935064935064935, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8831168831168831 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 154.0, "llamacpp:spec_decode_num_accepted_tokens_total": 140.0, "llamacpp:spec_decode_num_drafts_total": 77.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 72.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 68.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0034", "run": "context176-32768-cpu48-q8-u2048-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx32768-cpu48-q8_0-b8192-u2048", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 667.5658269858759, "decode_tps": 32.02160273824182, "first_frame_s": 12.273696751100942, "first_answer_s": null, "wall_s": 28.23195016104728, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:09:00.821636+00:00", "configuration_ids": [ "C007" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-u2048-control-20261005/requests.json", "bytes": 18057, "sha256": "466962bb262f626bd1930f939d96a9dd005892e3a93d77f30ff95dec62b558c0" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 12271.449, "prompt_per_token_ms": 1.4979796142578126, "prompt_per_second": 667.5658269858759, "predicted_n": 512, "predicted_ms": 15957.977, "predicted_per_token_ms": 31.22891780821918, "predicted_per_second": 32.02160273824182, "draft_n": 430, "draft_n_accepted": 296 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6883720930232559, "mean_accepted_span_including_target_bonus": 2.376744186046512, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7581395348837209, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6186046511627907 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 430.0, "llamacpp:spec_decode_num_accepted_tokens_total": 296.0, "llamacpp:spec_decode_num_drafts_total": 215.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 163.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 133.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0035", "run": "context176-32768-cpu48-q8-u2048-control-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx32768-cpu48-q8_0-b8192-u2048", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 218, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 715.8510142067415, "decode_tps": 35.646126284533615, "first_frame_s": 11.644156027003191, "first_answer_s": 16.492261828039773, "wall_s": 17.732033359003253, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T14:09:18.610694+00:00", "configuration_ids": [ "C007" ], "missing_configuration_match": false, "source": { "path": "context176-32768-cpu48-q8-u2048-control-20261005/requests.json", "bytes": 18057, "sha256": "466962bb262f626bd1930f939d96a9dd005892e3a93d77f30ff95dec62b558c0" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11443.722, "prompt_per_token_ms": 1.396938720703125, "prompt_per_second": 715.8510142067415, "predicted_n": 218, "predicted_ms": 6087.618, "predicted_per_token_ms": 28.053539170506916, "predicted_per_second": 35.646126284533615, "draft_n": 166, "draft_n_accepted": 136 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8192771084337349, "mean_accepted_span_including_target_bonus": 2.63855421686747, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8433734939759037, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7951807228915663 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 166.0, "llamacpp:spec_decode_num_accepted_tokens_total": 136.0, "llamacpp:spec_decode_num_drafts_total": 83.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 70.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 66.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0036", "run": "context176-40960-cpu44-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx40960-cpu44-f16-b8192-u8192", "context_tokens": 40960, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1040.9771664860755, "decode_tps": 33.197776827393064, "first_frame_s": 7.8721835860051215, "first_answer_s": null, "wall_s": 23.264837726950645, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 44, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:14:26.876396+00:00", "configuration_ids": [ "C008" ], "missing_configuration_match": false, "source": { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/requests.json", "bytes": 38009, "sha256": "19a64e38710db82a54606b61e0ab7264ed9d57f6e11337b5bb20873e2b9edf41" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7869.529, "prompt_per_token_ms": 0.9606358642578126, "prompt_per_second": 1040.9771664860755, "predicted_n": 512, "predicted_ms": 15392.597, "predicted_per_token_ms": 30.12249902152642, "predicted_per_second": 33.197776827393064, "draft_n": 435, "draft_n_accepted": 293 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6735632183908046, "mean_accepted_span_including_target_bonus": 2.3440366972477067, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7752293577981652, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5688073394495413 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 435.0, "llamacpp:spec_decode_num_accepted_tokens_total": 293.0, "llamacpp:spec_decode_num_drafts_total": 218.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 169.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0037", "run": "context176-40960-cpu44-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx40960-cpu44-f16-b8192-u8192", "context_tokens": 40960, "input_tokens": 36864, "output_tokens": 512, "fresh_input_tokens": 36864, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1033.9626030558272, "decode_tps": 28.029076519269196, "first_frame_s": 35.660386752919294, "first_answer_s": null, "wall_s": 53.89209714590106, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 44, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:15:20.884026+00:00", "configuration_ids": [ "C008" ], "missing_configuration_match": false, "source": { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/requests.json", "bytes": 38009, "sha256": "19a64e38710db82a54606b61e0ab7264ed9d57f6e11337b5bb20873e2b9edf41" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 36864, "prompt_ms": 35653.127, "prompt_per_token_ms": 0.9671529676649305, "prompt_per_second": 1033.9626030558272, "predicted_n": 512, "predicted_ms": 18231.068, "predicted_per_token_ms": 35.67723679060665, "predicted_per_second": 28.029076519269196, "draft_n": 466, "draft_n_accepted": 278 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 36864, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5965665236051502, "mean_accepted_span_including_target_bonus": 2.1931330472103, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6652360515021459, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5278969957081545 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 466.0, "llamacpp:spec_decode_num_accepted_tokens_total": 278.0, "llamacpp:spec_decode_num_drafts_total": 233.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 155.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 123.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "c631047aca77462df1c2cb6947effe7759d64de34efd46a318e734d1db1db283", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0038", "run": "context176-40960-cpu44-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx40960-cpu44-f16-b8192-u8192", "context_tokens": 40960, "input_tokens": 36864, "output_tokens": 249, "fresh_input_tokens": 36864, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1032.7795730858384, "decode_tps": 33.44201538845126, "first_frame_s": 36.31920716201421, "first_answer_s": 42.41083877708297, "wall_s": 43.735762527096085, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 44, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:16:04.727226+00:00", "configuration_ids": [ "C008" ], "missing_configuration_match": false, "source": { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/requests.json", "bytes": 38009, "sha256": "19a64e38710db82a54606b61e0ab7264ed9d57f6e11337b5bb20873e2b9edf41" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 36864, "prompt_ms": 35693.967, "prompt_per_token_ms": 0.9682608235677083, "prompt_per_second": 1032.7795730858384, "predicted_n": 249, "predicted_ms": 7415.821, "predicted_per_token_ms": 29.902504032258065, "predicted_per_second": 33.44201538845126, "draft_n": 190, "draft_n_accepted": 155 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 36864, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8157894736842105, "mean_accepted_span_including_target_bonus": 2.6315789473684212, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8526315789473684, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7789473684210526 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 190.0, "llamacpp:spec_decode_num_accepted_tokens_total": 155.0, "llamacpp:spec_decode_num_drafts_total": 95.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 81.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 74.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "61092d521b4333a400137baa939bc29fb24d51f0afd16c6d7df4f14894d9e657", "complete_native_template": true, "requested_output_budget": 4064, "truncated": false } }, { "id": "M0039", "run": "context176-45056-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx45056-cpu45-f16-b8192-u8192", "context_tokens": 45056, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1031.7232221099093, "decode_tps": 31.54761096301089, "first_frame_s": 7.942647886928171, "first_answer_s": null, "wall_s": 24.14044999796897, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 45, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:18:05.416035+00:00", "configuration_ids": [ "C009" ], "missing_configuration_match": false, "source": { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 39305, "sha256": "7eeab3be497b5af10eed75e335d98028459a1845c16df85606ce2d4635989a12" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7940.114, "prompt_per_token_ms": 0.969252197265625, "prompt_per_second": 1031.7232221099093, "predicted_n": 512, "predicted_ms": 16197.74, "predicted_per_token_ms": 31.69812133072407, "predicted_per_second": 31.54761096301089, "draft_n": 442, "draft_n_accepted": 289 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6538461538461539, "mean_accepted_span_including_target_bonus": 2.3076923076923075, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.751131221719457, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5565610859728507 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 442.0, "llamacpp:spec_decode_num_accepted_tokens_total": 289.0, "llamacpp:spec_decode_num_drafts_total": 221.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 123.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0040", "run": "context176-45056-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx45056-cpu45-f16-b8192-u8192", "context_tokens": 45056, "input_tokens": 40550, "output_tokens": 512, "fresh_input_tokens": 40550, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1027.0253814316548, "decode_tps": 30.81400593655041, "first_frame_s": 39.490885696024634, "first_answer_s": null, "wall_s": 56.0749419570202, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 45, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:19:01.615988+00:00", "configuration_ids": [ "C009" ], "missing_configuration_match": false, "source": { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 39305, "sha256": "7eeab3be497b5af10eed75e335d98028459a1845c16df85606ce2d4635989a12" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 40550, "prompt_ms": 39482.958, "prompt_per_token_ms": 0.9736857706535141, "prompt_per_second": 1027.0253814316548, "predicted_n": 512, "predicted_ms": 16583.368, "predicted_per_token_ms": 32.45277495107632, "predicted_per_second": 30.81400593655041, "draft_n": 426, "draft_n_accepted": 297 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 40550, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6971830985915493, "mean_accepted_span_including_target_bonus": 2.3943661971830985, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.755868544600939, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6384976525821596 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 426.0, "llamacpp:spec_decode_num_accepted_tokens_total": 297.0, "llamacpp:spec_decode_num_drafts_total": 213.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 161.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 136.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "366f1a575da5ec90fd6c4d1eba89f396609c221e3965a1af2687322d55afbe4b", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0041", "run": "context176-45056-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx45056-cpu45-f16-b8192-u8192", "context_tokens": 45056, "input_tokens": 40550, "output_tokens": 259, "fresh_input_tokens": 40550, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1027.0442924058673, "decode_tps": 34.87461561888047, "first_frame_s": 40.17556116206106, "first_answer_s": 46.26772591099143, "wall_s": 47.574317724094726, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 45, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:19:49.307510+00:00", "configuration_ids": [ "C009" ], "missing_configuration_match": false, "source": { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 39305, "sha256": "7eeab3be497b5af10eed75e335d98028459a1845c16df85606ce2d4635989a12" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 40550, "prompt_ms": 39482.231, "prompt_per_token_ms": 0.9736678421701603, "prompt_per_second": 1027.0442924058673, "predicted_n": 259, "predicted_ms": 7397.931, "predicted_per_token_ms": 28.674151162790697, "predicted_per_second": 34.87461561888047, "draft_n": 188, "draft_n_accepted": 166 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 40550, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8829787234042553, "mean_accepted_span_including_target_bonus": 2.7659574468085104, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.925531914893617, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8404255319148937 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 188.0, "llamacpp:spec_decode_num_accepted_tokens_total": 166.0, "llamacpp:spec_decode_num_drafts_total": 94.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 87.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 79.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7f34369bd5b2fb266da0a03985b03b82f1e928ea98bf40a44bebfeebc5b42ff2", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0042", "run": "context176-47104-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx47104-cpu45-f16-b8192-u8192", "context_tokens": 47104, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1032.7599994099955, "decode_tps": 31.798245545803194, "first_frame_s": 7.9346997699467465, "first_answer_s": null, "wall_s": 24.00484778592363, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 45, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:23:21.266722+00:00", "configuration_ids": [ "C010" ], "missing_configuration_match": false, "source": { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 41392, "sha256": "69db7db6d863e99cb279cfaa8b193e66adb751a440e496a6f101f1582c62dff6" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7932.143, "prompt_per_token_ms": 0.9682791748046875, "prompt_per_second": 1032.7599994099955, "predicted_n": 512, "predicted_ms": 16070.069, "predicted_per_token_ms": 31.4482759295499, "predicted_per_second": 31.798245545803194, "draft_n": 442, "draft_n_accepted": 289 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6538461538461539, "mean_accepted_span_including_target_bonus": 2.3076923076923075, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.751131221719457, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5565610859728507 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 442.0, "llamacpp:spec_decode_num_accepted_tokens_total": 289.0, "llamacpp:spec_decode_num_drafts_total": 221.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 123.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0043", "run": "context176-47104-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx47104-cpu45-f16-b8192-u8192", "context_tokens": 47104, "input_tokens": 42393, "output_tokens": 512, "fresh_input_tokens": 42393, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 995.0634823884114, "decode_tps": 28.125871365364095, "first_frame_s": 42.61133972601965, "first_answer_s": null, "wall_s": 60.780481619993225, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 45, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:24:22.168751+00:00", "configuration_ids": [ "C010" ], "missing_configuration_match": false, "source": { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 41392, "sha256": "69db7db6d863e99cb279cfaa8b193e66adb751a440e496a6f101f1582c62dff6" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 42393, "prompt_ms": 42603.312, "prompt_per_token_ms": 1.0049610077135376, "prompt_per_second": 995.0634823884114, "predicted_n": 512, "predicted_ms": 18168.326, "predicted_per_token_ms": 35.55445401174168, "predicted_per_second": 28.125871365364095, "draft_n": 457, "draft_n_accepted": 282 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 42393, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6170678336980306, "mean_accepted_span_including_target_bonus": 2.2314410480349345, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6943231441048034, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.537117903930131 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 457.0, "llamacpp:spec_decode_num_accepted_tokens_total": 282.0, "llamacpp:spec_decode_num_drafts_total": 229.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 123.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "4b27c982d1eb67aae1fb35b49d495f57ee3f31d4e02e12b2552d9588f156bef1", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0044", "run": "context176-47104-cpu45-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx47104-cpu45-f16-b8192-u8192", "context_tokens": 47104, "input_tokens": 42393, "output_tokens": 283, "fresh_input_tokens": 42393, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 992.4323899142855, "decode_tps": 33.017195636484885, "first_frame_s": 43.41855191800278, "first_answer_s": 50.68830344895832, "wall_s": 51.960397689952515, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 45, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:25:14.256724+00:00", "configuration_ids": [ "C010" ], "missing_configuration_match": false, "source": { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 41392, "sha256": "69db7db6d863e99cb279cfaa8b193e66adb751a440e496a6f101f1582c62dff6" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 42393, "prompt_ms": 42716.26, "prompt_per_token_ms": 1.0076253155002006, "prompt_per_second": 992.4323899142855, "predicted_n": 283, "predicted_ms": 8541.004, "predicted_per_token_ms": 30.287248226950357, "predicted_per_second": 33.017195636484885, "draft_n": 216, "draft_n_accepted": 174 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 42393, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8055555555555556, "mean_accepted_span_including_target_bonus": 2.611111111111111, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8333333333333334, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7777777777777778 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 216.0, "llamacpp:spec_decode_num_accepted_tokens_total": 174.0, "llamacpp:spec_decode_num_drafts_total": 108.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 90.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 84.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "079160d9cf4bac72cccd96b3f955d63993d57c7cc72bafd339ee89a00091f8eb", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0045", "run": "context176-49152-cpu46-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx49152-cpu46-f16-b8192-u8192", "context_tokens": 49152, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1028.9616530912058, "decode_tps": 30.271037749228086, "first_frame_s": 7.963711199001409, "first_answer_s": null, "wall_s": 24.84462779108435, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 46, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:09:17.807654+00:00", "configuration_ids": [ "C011" ], "missing_configuration_match": false, "source": { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/requests.json", "bytes": 42225, "sha256": "bf6bf41fc09f376f8f2fdf312f21a55d62d1781c6ae80d2f904c31a6662413f5" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7961.424, "prompt_per_token_ms": 0.971853515625, "prompt_per_second": 1028.9616530912058, "predicted_n": 512, "predicted_ms": 16880.822, "predicted_per_token_ms": 33.03487671232877, "predicted_per_second": 30.271037749228086, "draft_n": 457, "draft_n_accepted": 281 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6148796498905909, "mean_accepted_span_including_target_bonus": 2.2270742358078603, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6899563318777293, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.537117903930131 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 457.0, "llamacpp:spec_decode_num_accepted_tokens_total": 281.0, "llamacpp:spec_decode_num_drafts_total": 229.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 158.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 123.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0046", "run": "context176-49152-cpu46-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx49152-cpu46-f16-b8192-u8192", "context_tokens": 49152, "input_tokens": 44236, "output_tokens": 512, "fresh_input_tokens": 44236, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 989.9301306615572, "decode_tps": 28.182926496335533, "first_frame_s": 44.694684657035396, "first_answer_s": null, "wall_s": 62.82718886900693, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 46, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:10:20.756229+00:00", "configuration_ids": [ "C011" ], "missing_configuration_match": false, "source": { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/requests.json", "bytes": 42225, "sha256": "bf6bf41fc09f376f8f2fdf312f21a55d62d1781c6ae80d2f904c31a6662413f5" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 44236, "prompt_ms": 44685.982, "prompt_per_token_ms": 1.0101723031015464, "prompt_per_second": 989.9301306615572, "predicted_n": 512, "predicted_ms": 18131.545, "predicted_per_token_ms": 35.482475538160465, "predicted_per_second": 28.182926496335533, "draft_n": 454, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 44236, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6233480176211453, "mean_accepted_span_including_target_bonus": 2.246696035242291, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7092511013215859, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5374449339207048 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 454.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 161.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 122.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "d747f86c4a7b0c6e7ef8b5e3fdd24d536818008ae791aa6e9fe59ec3cccd315a", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0047", "run": "context176-49152-cpu46-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx49152-cpu46-f16-b8192-u8192", "context_tokens": 49152, "input_tokens": 44236, "output_tokens": 179, "fresh_input_tokens": 44236, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 989.8347710320143, "decode_tps": 30.348185758176104, "first_frame_s": 45.44891640602145, "first_answer_s": 49.935837459983304, "wall_s": 51.31503222999163, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 46, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:11:12.193442+00:00", "configuration_ids": [ "C011" ], "missing_configuration_match": false, "source": { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/requests.json", "bytes": 42225, "sha256": "bf6bf41fc09f376f8f2fdf312f21a55d62d1781c6ae80d2f904c31a6662413f5" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 44236, "prompt_ms": 44690.287, "prompt_per_token_ms": 1.010269622027308, "prompt_per_second": 989.8347710320143, "predicted_n": 179, "predicted_ms": 5865.26, "predicted_per_token_ms": 32.950898876404494, "predicted_per_second": 30.348185758176104, "draft_n": 146, "draft_n_accepted": 107 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 44236, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7328767123287672, "mean_accepted_span_including_target_bonus": 2.4657534246575343, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7808219178082192, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.684931506849315 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 146.0, "llamacpp:spec_decode_num_accepted_tokens_total": 107.0, "llamacpp:spec_decode_num_drafts_total": 73.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 57.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 50.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "18209914819d483a1130452bc9fbea1e43e064b5c1858701dea7a8b6f81a5ab1", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0048", "run": "context176-65536-cpu48-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx65536-cpu48-f16-b8192-u8192", "context_tokens": 65536, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1019.9571448670333, "decode_tps": 32.834208387476046, "first_frame_s": 8.033985801041126, "first_answer_s": null, "wall_s": 23.597181156044826, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:03:38.557041+00:00", "configuration_ids": [ "C012" ], "missing_configuration_match": false, "source": { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/requests.json", "bytes": 50549, "sha256": "dc983c4ef8356d366d761ce45d9b5e990f887f0c425068b76374a43b05ff494c" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8031.71, "prompt_per_token_ms": 0.980433349609375, "prompt_per_second": 1019.9571448670333, "predicted_n": 512, "predicted_ms": 15563.037, "predicted_per_token_ms": 30.456041095890413, "predicted_per_second": 32.834208387476046, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7990430622009569, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.645933014354067 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 167.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 135.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0049", "run": "context176-65536-cpu48-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ctx65536-cpu48-f16-b8192-u8192", "context_tokens": 65536, "input_tokens": 58982, "output_tokens": 512, "fresh_input_tokens": 58982, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 920.8173714659114, "decode_tps": 28.26777091311649, "first_frame_s": 64.06437252392061, "first_answer_s": null, "wall_s": 82.14301804697607, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:05:00.849497+00:00", "configuration_ids": [ "C012" ], "missing_configuration_match": false, "source": { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/requests.json", "bytes": 50549, "sha256": "dc983c4ef8356d366d761ce45d9b5e990f887f0c425068b76374a43b05ff494c" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 58982, "prompt_ms": 64053.961, "prompt_per_token_ms": 1.0859916754264014, "prompt_per_second": 920.8173714659114, "predicted_n": 512, "predicted_ms": 18077.124, "predicted_per_token_ms": 35.37597651663405, "predicted_per_second": 28.26777091311649, "draft_n": 424, "draft_n_accepted": 299 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 58982, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7051886792452831, "mean_accepted_span_including_target_bonus": 2.410377358490566, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7405660377358491, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6698113207547169 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 424.0, "llamacpp:spec_decode_num_accepted_tokens_total": 299.0, "llamacpp:spec_decode_num_drafts_total": 212.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 157.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 142.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "44c3cd0ced7cf161cfc7513955726a186c34a65ddd043ad4c486698487bed42e", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0050", "run": "context176-65536-cpu48-f16-u8192-screen-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ctx65536-cpu48-f16-b8192-u8192", "context_tokens": 65536, "input_tokens": 58982, "output_tokens": 336, "fresh_input_tokens": 58982, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 920.7032860624892, "decode_tps": 28.910823142052422, "first_frame_s": 64.99529734905809, "first_answer_s": 75.10889561707154, "wall_s": 76.5840275370283, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_CONTEXT_PROTOCOL_MATRIX_NOT_PRODUCTION_ACCEPTANCE", "threads": 14, "cpu_moe": 48, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T15:06:17.582585+00:00", "configuration_ids": [ "C012" ], "missing_configuration_match": false, "source": { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/requests.json", "bytes": 50549, "sha256": "dc983c4ef8356d366d761ce45d9b5e990f887f0c425068b76374a43b05ff494c" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 58982, "prompt_ms": 64061.898, "prompt_per_token_ms": 1.0861262419043098, "prompt_per_second": 920.7032860624892, "predicted_n": 336, "predicted_ms": 11587.356, "predicted_per_token_ms": 34.5891223880597, "predicted_per_second": 28.910823142052422, "draft_n": 266, "draft_n_accepted": 204 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 58982, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7669172932330827, "mean_accepted_span_including_target_bonus": 2.533834586466165, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8120300751879699, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7218045112781954 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 266.0, "llamacpp:spec_decode_num_accepted_tokens_total": 204.0, "llamacpp:spec_decode_num_drafts_total": 133.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 108.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 96.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "1b5f6470130c47240bbd6868f4be8c3e44f9ba118eb0e691c279885473579a5d", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0051", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "baseline96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 582.2138225748861, "decode_tps": 31.970071508478735, "first_frame_s": 14.073027804028243, "first_answer_s": null, "wall_s": 30.056856779032387, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:17:41.249468+00:00", "configuration_ids": [ "C013" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14070.432, "prompt_per_token_ms": 1.71758203125, "prompt_per_second": 582.2138225748861, "predicted_n": 512, "predicted_ms": 15983.699, "predicted_per_token_ms": 31.279254403131116, "predicted_per_second": 31.970071508478735 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0052", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "baseline96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.119985427206416, "decode_tps": 32.1509400437819, "first_frame_s": 0.12300086708273739, "first_answer_s": null, "wall_s": 16.01691461307928, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:17:57.307918+00:00", "configuration_ids": [ "C013" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 120.773, "prompt_per_token_ms": 30.19325, "prompt_per_second": 33.119985427206416, "predicted_n": 512, "predicted_ms": 15893.781, "predicted_per_token_ms": 31.10328962818004, "predicted_per_second": 32.1509400437819 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0053", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "baseline96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 180, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 594.2940085341257, "decode_tps": 33.052442963301075, "first_frame_s": 13.965783815016039, "first_answer_s": 17.658152548014186, "wall_s": 19.38156972604338, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:18:16.747511+00:00", "configuration_ids": [ "C013" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 13784.423, "prompt_per_token_ms": 1.6826688232421876, "prompt_per_second": 594.2940085341257, "predicted_n": 180, "predicted_ms": 5415.636, "predicted_per_token_ms": 30.254949720670393, "predicted_per_second": 33.052442963301075 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0054", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 722.2901181595039, "decode_tps": 30.618673364186886, "first_frame_s": 11.344250094029121, "first_answer_s": null, "wall_s": 28.03353988903109, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:18:50.330107+00:00", "configuration_ids": [ "C014" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11341.703, "prompt_per_token_ms": 1.3844852294921874, "prompt_per_second": 722.2901181595039, "predicted_n": 512, "predicted_ms": 16689.162, "predicted_per_token_ms": 32.65980821917808, "predicted_per_second": 30.618673364186886 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0055", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.19915342158775, "decode_tps": 30.680712549041857, "first_frame_s": 0.12299151392653584, "first_answer_s": null, "wall_s": 16.778580172918737, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:19:07.149551+00:00", "configuration_ids": [ "C014" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 120.485, "prompt_per_token_ms": 30.12125, "prompt_per_second": 33.19915342158775, "predicted_n": 512, "predicted_ms": 16655.415, "predicted_per_token_ms": 32.59376712328767, "predicted_per_second": 30.680712549041857 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0056", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 162, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 744.7714173603745, "decode_tps": 32.88043428314219, "first_frame_s": 11.184126296895556, "first_answer_s": 14.37217122700531, "wall_s": 16.080769914900884, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:19:23.290050+00:00", "configuration_ids": [ "C014" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10999.348, "prompt_per_token_ms": 1.34269384765625, "prompt_per_second": 744.7714173603745, "predicted_n": 162, "predicted_ms": 4896.529, "predicted_per_token_ms": 30.413223602484475, "predicted_per_second": 32.88043428314219 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0057", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "overlap-cache112", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 580.622709350563, "decode_tps": 26.774397200121115, "first_frame_s": 14.11131970200222, "first_answer_s": null, "wall_s": 33.19691417901777, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 112, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:20:02.178728+00:00", "configuration_ids": [ "C016" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14108.99, "prompt_per_token_ms": 1.722288818359375, "prompt_per_second": 580.622709350563, "predicted_n": 512, "predicted_ms": 19085.397, "predicted_per_token_ms": 37.349113502935424, "predicted_per_second": 26.774397200121115 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0058", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "overlap-cache112", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.49205975289789, "decode_tps": 26.85464850549677, "first_frame_s": 0.12512556393630803, "first_answer_s": null, "wall_s": 19.153710733982734, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 112, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:20:21.374140+00:00", "configuration_ids": [ "C016" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 123.107, "prompt_per_token_ms": 30.77675, "prompt_per_second": 32.49205975289789, "predicted_n": 512, "predicted_ms": 19028.363, "predicted_per_token_ms": 37.237500978473584, "predicted_per_second": 26.85464850549677 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0059", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "overlap-cache112", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 203, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 592.8095358392513, "decode_tps": 27.24675824275135, "first_frame_s": 14.006583124981262, "first_answer_s": 19.34478250495158, "wall_s": 21.420470827957615, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 112, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:20:42.854137+00:00", "configuration_ids": [ "C016" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 13818.941, "prompt_per_token_ms": 1.6868824462890626, "prompt_per_second": 592.8095358392513, "predicted_n": 203, "predicted_ms": 7413.726, "predicted_per_token_ms": 36.701613861386136, "predicted_per_second": 27.24675824275135 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0060", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 550.104574933416, "decode_tps": 36.48682304447765, "first_frame_s": 14.894142839009874, "first_answer_s": null, "wall_s": 28.899442695081234, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:21:17.874988+00:00", "configuration_ids": [ "C015" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14891.714, "prompt_per_token_ms": 1.817836181640625, "prompt_per_second": 550.104574933416, "predicted_n": 512, "predicted_ms": 14005.056, "predicted_per_token_ms": 27.407154598825834, "predicted_per_second": 36.48682304447765, "draft_n": 422, "draft_n_accepted": 299 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7085308056872038, "mean_accepted_span_including_target_bonus": 2.4170616113744074, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7819905213270142, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6350710900473934 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 422.0, "llamacpp:spec_decode_num_accepted_tokens_total": 299.0, "llamacpp:spec_decode_num_drafts_total": 211.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 134.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0061", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.978265757325936, "decode_tps": 38.03438874420274, "first_frame_s": 0.1355082190129906, "first_answer_s": null, "wall_s": 13.570949633023702, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:21:31.486594+00:00", "configuration_ids": [ "C015" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 133.43, "prompt_per_token_ms": 33.3575, "prompt_per_second": 29.978265757325936, "predicted_n": 512, "predicted_ms": 13435.21, "predicted_per_token_ms": 26.291996086105673, "predicted_per_second": 38.03438874420274, "draft_n": 422, "draft_n_accepted": 299 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7085308056872038, "mean_accepted_span_including_target_bonus": 2.4170616113744074, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7819905213270142, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6350710900473934 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 422.0, "llamacpp:spec_decode_num_accepted_tokens_total": 299.0, "llamacpp:spec_decode_num_drafts_total": 211.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 134.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0062", "run": "guided-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "gpu-mtp2-cache96", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 271, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 568.3986056194167, "decode_tps": 41.42554188444483, "first_frame_s": 14.616702980012633, "first_answer_s": 19.836100685060956, "wall_s": 21.13468933501281, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 96, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:21:52.681082+00:00", "configuration_ids": [ "C015" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14412.421, "prompt_per_token_ms": 1.7593287353515625, "prompt_per_second": 568.3986056194167, "predicted_n": 271, "predicted_ms": 6517.718, "predicted_per_token_ms": 24.139696296296297, "predicted_per_second": 41.42554188444483, "draft_n": 210, "draft_n_accepted": 166 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7904761904761904, "mean_accepted_span_including_target_bonus": 2.5809523809523807, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8571428571428571, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7238095238095238 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 210.0, "llamacpp:spec_decode_num_accepted_tokens_total": 166.0, "llamacpp:spec_decode_num_drafts_total": 105.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 90.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 76.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0063", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 721.5347834021634, "decode_tps": 31.1324821148155, "first_frame_s": 11.35594170098193, "first_answer_s": null, "wall_s": 27.7698010050226, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:22:30.166076+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11353.576, "prompt_per_token_ms": 1.3859345703125, "prompt_per_second": 721.5347834021634, "predicted_n": 512, "predicted_ms": 16413.725, "predicted_per_token_ms": 32.12079256360078, "predicted_per_second": 31.1324821148155 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0064", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.972375867250694, "decode_tps": 31.339016836779468, "first_frame_s": 0.12723471596837044, "first_answer_s": null, "wall_s": 16.432893012068234, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:22:46.641227+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 125.108, "prompt_per_token_ms": 31.277, "prompt_per_second": 31.972375867250694, "predicted_n": 512, "predicted_ms": 16305.553, "predicted_per_token_ms": 31.909105675146773, "predicted_per_second": 31.339016836779468 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0065", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 748.722866336334, "decode_tps": 31.266763436585652, "first_frame_s": 10.943594697047956, "first_answer_s": null, "wall_s": 27.28695606905967, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:23:13.971257+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10941.298, "prompt_per_token_ms": 1.335607666015625, "prompt_per_second": 748.722866336334, "predicted_n": 512, "predicted_ms": 16343.233, "predicted_per_token_ms": 31.982843444227008, "predicted_per_second": 31.266763436585652 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0066", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.55340793489319, "decode_tps": 31.26082428981953, "first_frame_s": 0.12521703692618757, "first_answer_s": null, "wall_s": 16.471789696021006, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:23:30.485832+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 122.875, "prompt_per_token_ms": 30.71875, "prompt_per_second": 32.55340793489319, "predicted_n": 512, "predicted_ms": 16346.338, "predicted_per_token_ms": 31.98891976516634, "predicted_per_second": 31.26082428981953 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0067", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 742.6534467342422, "decode_tps": 31.2803104524902, "first_frame_s": 11.033200309961103, "first_answer_s": null, "wall_s": 27.369481033994816, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:23:57.897796+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11030.717, "prompt_per_token_ms": 1.3465230712890626, "prompt_per_second": 742.6534467342422, "predicted_n": 512, "predicted_ms": 16336.155, "predicted_per_token_ms": 31.96899217221135, "predicted_per_second": 31.2803104524902 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0068", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.241089310496704, "decode_tps": 31.216408761706763, "first_frame_s": 0.12268548202700913, "first_answer_s": null, "wall_s": 16.492386016063392, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:24:14.434360+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 120.333, "prompt_per_token_ms": 30.08325, "prompt_per_second": 33.241089310496704, "predicted_n": 512, "predicted_ms": 16369.596, "predicted_per_token_ms": 32.034434442270054, "predicted_per_second": 31.216408761706763 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0069", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 742.5669429719103, "decode_tps": 31.13187896485954, "first_frame_s": 11.034177948022261, "first_answer_s": null, "wall_s": 27.44834012107458, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:24:41.925971+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11032.002, "prompt_per_token_ms": 1.346679931640625, "prompt_per_second": 742.5669429719103, "predicted_n": 512, "predicted_ms": 16414.043, "predicted_per_token_ms": 32.12141487279844, "predicted_per_second": 31.13187896485954 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0070", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.150981006807974, "decode_tps": 31.197453506331254, "first_frame_s": 0.12696017196867615, "first_answer_s": null, "wall_s": 16.50664922001306, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:24:58.474773+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 124.413, "prompt_per_token_ms": 31.10325, "prompt_per_second": 32.150981006807974, "predicted_n": 512, "predicted_ms": 16379.542, "predicted_per_token_ms": 32.05389823874755, "predicted_per_second": 31.197453506331254 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0071", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 740.976248871056, "decode_tps": 31.145525731903096, "first_frame_s": 11.058148019015789, "first_answer_s": null, "wall_s": 27.465182592975907, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:25:25.982615+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11055.685, "prompt_per_token_ms": 1.3495709228515624, "prompt_per_second": 740.976248871056, "predicted_n": 512, "predicted_ms": 16406.851, "predicted_per_token_ms": 32.10734050880626, "predicted_per_second": 31.145525731903096 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0072", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.93995288856949, "decode_tps": 31.051316767243158, "first_frame_s": 0.1274781080428511, "first_answer_s": null, "wall_s": 16.58431752503384, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:25:42.609671+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 125.235, "prompt_per_token_ms": 31.30875, "prompt_per_second": 31.93995288856949, "predicted_n": 512, "predicted_ms": 16456.629, "predicted_per_token_ms": 32.20475342465753, "predicted_per_second": 31.051316767243158 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0073", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 229, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 733.9834893137846, "decode_tps": 31.849985450982523, "first_frame_s": 11.348060077056289, "first_answer_s": 16.792262290022336, "wall_s": 18.50675665808376, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:26:01.177868+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11161.014, "prompt_per_token_ms": 1.362428466796875, "prompt_per_second": 733.9834893137846, "predicted_n": 229, "predicted_ms": 7158.559, "predicted_per_token_ms": 31.39718859649123, "predicted_per_second": 31.849985450982523 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0074", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 214, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 721.847655074544, "decode_tps": 30.214677410797904, "first_frame_s": 22.84231852099765, "first_answer_s": 28.10675168596208, "wall_s": 29.89220966002904, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:26:31.142768+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22697.31, "prompt_per_token_ms": 1.3853338623046876, "prompt_per_second": 721.847655074544, "predicted_n": 214, "predicted_ms": 7049.554, "predicted_per_token_ms": 33.09649765258216, "predicted_per_second": 30.214677410797904 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0075", "run": "guided-speed175-finalist-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 266, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 678.7577974763997, "decode_tps": 27.81439131300931, "first_frame_s": 42.48376567999367, "first_answer_s": 50.054293968947604, "wall_s": 52.011795619968325, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:27:23.256706+00:00", "configuration_ids": [ "C017" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 42241.872, "prompt_per_token_ms": 1.4732795758928572, "prompt_per_second": 678.7577974763997, "predicted_n": 266, "predicted_ms": 9527.442, "predicted_per_token_ms": 35.95261132075471, "predicted_per_second": 27.81439131300931 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0076", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 663.5308784850761, "decode_tps": 30.832957474955766, "first_frame_s": 12.34829370805528, "first_answer_s": null, "wall_s": 28.921561789000407, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:42:59.680997+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 12346.072, "prompt_per_token_ms": 1.5070888671875, "prompt_per_second": 663.5308784850761, "predicted_n": 512, "predicted_ms": 16573.175, "predicted_per_token_ms": 32.4328277886497, "predicted_per_second": 30.832957474955766 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0077", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 30.57613074353506, "decode_tps": 30.793761943128562, "first_frame_s": 0.13300402194727212, "first_answer_s": null, "wall_s": 16.727403554017656, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:43:16.449876+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 130.821, "prompt_per_token_ms": 32.70525, "prompt_per_second": 30.57613074353506, "predicted_n": 512, "predicted_ms": 16594.27, "predicted_per_token_ms": 32.4741095890411, "predicted_per_second": 30.793761943128562 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0078", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 746.4371566794554, "decode_tps": 30.84791687216853, "first_frame_s": 10.977260002982803, "first_answer_s": null, "wall_s": 27.54249968705699, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:43:44.034223+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10974.802, "prompt_per_token_ms": 1.339697509765625, "prompt_per_second": 746.4371566794554, "predicted_n": 512, "predicted_ms": 16565.138, "predicted_per_token_ms": 32.41709980430528, "predicted_per_second": 30.84791687216853 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0079", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.14968895175939, "decode_tps": 30.889851384149452, "first_frame_s": 0.1270155319944024, "first_answer_s": null, "wall_s": 16.66980812093243, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:44:00.744861+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 124.418, "prompt_per_token_ms": 31.1045, "prompt_per_second": 32.14968895175939, "predicted_n": 512, "predicted_ms": 16542.65, "predicted_per_token_ms": 32.37309197651663, "predicted_per_second": 30.889851384149452 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0080", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 749.0839449511084, "decode_tps": 30.833662587773365, "first_frame_s": 10.938565628952347, "first_answer_s": null, "wall_s": 27.511468100943603, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:44:28.299037+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10936.024, "prompt_per_token_ms": 1.3349638671875, "prompt_per_second": 749.0839449511084, "predicted_n": 512, "predicted_ms": 16572.796, "predicted_per_token_ms": 32.43208610567515, "predicted_per_second": 30.833662587773365 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0081", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.63228311768833, "decode_tps": 30.71825594366692, "first_frame_s": 0.1247946850489825, "first_answer_s": null, "wall_s": 16.760070328018628, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:44:45.102717+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 122.578, "prompt_per_token_ms": 30.6445, "prompt_per_second": 32.63228311768833, "predicted_n": 512, "predicted_ms": 16635.059, "predicted_per_token_ms": 32.55393150684932, "predicted_per_second": 30.71825594366692 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0082", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 250, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 730.1420220709313, "decode_tps": 31.19638512826976, "first_frame_s": 11.440304767922498, "first_answer_s": 17.674332161899656, "wall_s": 19.42211080098059, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:45:04.581479+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11219.735, "prompt_per_token_ms": 1.3695965576171876, "prompt_per_second": 730.1420220709313, "predicted_n": 250, "predicted_ms": 7981.694, "predicted_per_token_ms": 32.05499598393575, "predicted_per_second": 31.19638512826976 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0083", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 184, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 729.4928751111227, "decode_tps": 30.276589049404443, "first_frame_s": 22.599783696001396, "first_answer_s": 27.32811846304685, "wall_s": 28.644369682995602, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:45:33.300529+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22459.438, "prompt_per_token_ms": 1.3708153076171874, "prompt_per_second": 729.4928751111227, "predicted_n": 184, "predicted_ms": 6044.274, "predicted_per_token_ms": 33.02881967213115, "predicted_per_second": 30.276589049404443 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0084", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 181, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 683.8053817191146, "decode_tps": 28.365208999587445, "first_frame_s": 42.165894908015616, "first_answer_s": 46.55670915497467, "wall_s": 48.512296368018724, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:46:21.909760+00:00", "configuration_ids": [ "C018" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 41930.059, "prompt_per_token_ms": 1.4624044015066964, "prompt_per_second": 683.8053817191146, "predicted_n": 181, "predicted_ms": 6345.802, "predicted_per_token_ms": 35.25445555555555, "predicted_per_second": 28.365208999587445 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0085", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 542.6543761279769, "decode_tps": 33.86723381580334, "first_frame_s": 15.098430631915107, "first_answer_s": null, "wall_s": 30.18702219298575, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:46:58.380805+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 15096.165, "prompt_per_token_ms": 1.8427935791015626, "prompt_per_second": 542.6543761279769, "predicted_n": 512, "predicted_ms": 15088.33, "predicted_per_token_ms": 29.52706457925636, "predicted_per_second": 33.86723381580334, "draft_n": 404, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7623762376237624, "mean_accepted_span_including_target_bonus": 2.5247524752475248, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8118811881188119, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7128712871287128 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 404.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 202.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 144.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0086", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.455081001472752, "decode_tps": 36.37162634481597, "first_frame_s": 0.1378594939596951, "first_answer_s": null, "wall_s": 14.187523848027922, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:47:12.609718+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 135.8, "prompt_per_token_ms": 33.95, "prompt_per_second": 29.455081001472752, "predicted_n": 512, "predicted_ms": 14049.413, "predicted_per_token_ms": 27.49395890410959, "predicted_per_second": 36.37162634481597, "draft_n": 404, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7623762376237624, "mean_accepted_span_including_target_bonus": 2.5247524752475248, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8118811881188119, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7128712871287128 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 404.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 202.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 144.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0087", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 565.9012447064194, "decode_tps": 36.40874148945667, "first_frame_s": 14.478245838079602, "first_answer_s": null, "wall_s": 28.513571181101725, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:47:41.166677+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14476.024, "prompt_per_token_ms": 1.7670927734375, "prompt_per_second": 565.9012447064194, "predicted_n": 512, "predicted_ms": 14035.091, "predicted_per_token_ms": 27.465931506849316, "predicted_per_second": 36.40874148945667, "draft_n": 404, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7623762376237624, "mean_accepted_span_including_target_bonus": 2.5247524752475248, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8118811881188119, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7128712871287128 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 404.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 202.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 144.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0088", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.486782549722086, "decode_tps": 36.04524800188705, "first_frame_s": 0.13763182295951992, "first_answer_s": null, "wall_s": 14.314502842025831, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:47:55.523442+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 135.654, "prompt_per_token_ms": 33.9135, "prompt_per_second": 29.486782549722086, "predicted_n": 512, "predicted_ms": 14176.626, "predicted_per_token_ms": 27.742908023483366, "predicted_per_second": 36.04524800188705, "draft_n": 404, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7623762376237624, "mean_accepted_span_including_target_bonus": 2.5247524752475248, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8118811881188119, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7128712871287128 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 404.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 202.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 144.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0089", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 562.7416399000309, "decode_tps": 36.174186850647686, "first_frame_s": 14.559672396979295, "first_answer_s": null, "wall_s": 28.686030827928334, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:48:24.251496+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14557.302, "prompt_per_token_ms": 1.777014404296875, "prompt_per_second": 562.7416399000309, "predicted_n": 512, "predicted_ms": 14126.095, "predicted_per_token_ms": 27.644021526418786, "predicted_per_second": 36.174186850647686, "draft_n": 404, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7623762376237624, "mean_accepted_span_including_target_bonus": 2.5247524752475248, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8118811881188119, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7128712871287128 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 404.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 202.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 144.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0090", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.226460895766735, "decode_tps": 36.06956131366613, "first_frame_s": 0.14369626890402287, "first_answer_s": null, "wall_s": 14.311014733975753, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:48:38.606451+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 141.711, "prompt_per_token_ms": 35.42775, "prompt_per_second": 28.226460895766735, "predicted_n": 512, "predicted_ms": 14167.07, "predicted_per_token_ms": 27.724207436399215, "predicted_per_second": 36.06956131366613, "draft_n": 404, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7623762376237624, "mean_accepted_span_including_target_bonus": 2.5247524752475248, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8118811881188119, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7128712871287128 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 404.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 202.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 164.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 144.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0091", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 197, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 559.8889462462187, "decode_tps": 41.31120935403812, "first_frame_s": 14.836953897029161, "first_answer_s": 18.372227676096372, "wall_s": 19.581763730035163, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:48:58.246513+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14631.473, "prompt_per_token_ms": 1.7860684814453125, "prompt_per_second": 559.8889462462187, "predicted_n": 197, "predicted_ms": 4744.475, "predicted_per_token_ms": 24.20650510204082, "predicted_per_second": 41.31120935403812, "draft_n": 142, "draft_n_accepted": 125 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8802816901408451, "mean_accepted_span_including_target_bonus": 2.76056338028169, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9154929577464789, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8450704225352113 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 142.0, "llamacpp:spec_decode_num_accepted_tokens_total": 125.0, "llamacpp:spec_decode_num_drafts_total": 71.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 60.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0092", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 201, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 554.927942349273, "decode_tps": 39.999704002190384, "first_frame_s": 29.723210864933208, "first_answer_s": 33.40038581995759, "wall_s": 34.723707247991115, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:49:33.042034+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 16, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 29524.554, "prompt_per_token_ms": 1.8020357666015625, "prompt_per_second": 554.927942349273, "predicted_n": 201, "predicted_ms": 5000.037, "predicted_per_token_ms": 25.000185000000002, "predicted_per_second": 39.999704002190384, "draft_n": 144, "draft_n_accepted": 129 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8958333333333334, "mean_accepted_span_including_target_bonus": 2.791666666666667, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9166666666666666, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.875 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 144.0, "llamacpp:spec_decode_num_accepted_tokens_total": 129.0, "llamacpp:spec_decode_num_drafts_total": 72.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 66.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 63.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0093", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "gpu-mtp2-cache64", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 280, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 541.3339490302141, "decode_tps": 34.60222203315711, "first_frame_s": 53.24895716004539, "first_answer_s": 59.878958277986385, "wall_s": 61.31284925097134, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 4096, "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 64, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:50:34.451610+00:00", "configuration_ids": [ "C019" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 17, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 52965.457, "prompt_per_token_ms": 1.847288539341518, "prompt_per_second": 541.3339490302141, "predicted_n": 280, "predicted_ms": 8063.066, "predicted_per_token_ms": 28.899878136200716, "predicted_per_second": 34.60222203315711, "draft_n": 222, "draft_n_accepted": 169 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7612612612612613, "mean_accepted_span_including_target_bonus": 2.5225225225225225, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8108108108108109, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7117117117117117 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 222.0, "llamacpp:spec_decode_num_accepted_tokens_total": 169.0, "llamacpp:spec_decode_num_drafts_total": 111.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 90.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 79.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0094", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 617.742310212965, "decode_tps": 28.34371373315054, "first_frame_s": 13.263563175918534, "first_answer_s": null, "wall_s": 31.292489203973673, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:51:13.041576+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 18, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 13261.193, "prompt_per_token_ms": 1.6187979736328124, "prompt_per_second": 617.742310212965, "predicted_n": 512, "predicted_ms": 18028.689, "predicted_per_token_ms": 35.28119178082191, "predicted_per_second": 28.34371373315054, "draft_n": 408, "draft_n_accepted": 307 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7524509803921569, "mean_accepted_span_including_target_bonus": 2.5049019607843137, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8137254901960784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6911764705882353 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 408.0, "llamacpp:spec_decode_num_accepted_tokens_total": 307.0, "llamacpp:spec_decode_num_drafts_total": 204.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 141.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0095", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.683193533205994, "decode_tps": 29.31069392924479, "first_frame_s": 0.1469153050566092, "first_answer_s": null, "wall_s": 17.581206753035076, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:51:30.667929+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 19, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 144.492, "prompt_per_token_ms": 36.123, "prompt_per_second": 27.683193533205994, "predicted_n": 512, "predicted_ms": 17433.91, "predicted_per_token_ms": 34.11724070450098, "predicted_per_second": 29.31069392924479, "draft_n": 408, "draft_n_accepted": 307 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7524509803921569, "mean_accepted_span_including_target_bonus": 2.5049019607843137, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8137254901960784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6911764705882353 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 408.0, "llamacpp:spec_decode_num_accepted_tokens_total": 307.0, "llamacpp:spec_decode_num_drafts_total": 204.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 141.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0096", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 675.0750538918711, "decode_tps": 29.5254599890438, "first_frame_s": 12.137349101016298, "first_answer_s": null, "wall_s": 29.444807959022, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:52:00.154845+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 20, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 12134.947, "prompt_per_token_ms": 1.4813167724609375, "prompt_per_second": 675.0750538918711, "predicted_n": 512, "predicted_ms": 17307.097, "predicted_per_token_ms": 33.869074363992176, "predicted_per_second": 29.5254599890438, "draft_n": 406, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7586206896551724, "mean_accepted_span_including_target_bonus": 2.5172413793103448, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8177339901477833, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6995073891625616 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 406.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 203.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 142.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0097", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.113183677485555, "decode_tps": 29.677609399450635, "first_frame_s": 0.144311262993142, "first_answer_s": null, "wall_s": 17.3630022269208, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:52:17.560563+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 21, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 142.282, "prompt_per_token_ms": 35.5705, "prompt_per_second": 28.113183677485555, "predicted_n": 512, "predicted_ms": 17218.368, "predicted_per_token_ms": 33.69543639921722, "predicted_per_second": 29.677609399450635, "draft_n": 406, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7586206896551724, "mean_accepted_span_including_target_bonus": 2.5172413793103448, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8177339901477833, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6995073891625616 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 406.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 203.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 142.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0098", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 674.4256905474448, "decode_tps": 29.802384529028135, "first_frame_s": 12.148791228886694, "first_answer_s": null, "wall_s": 29.295278781908564, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:52:46.900117+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 22, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 12146.631, "prompt_per_token_ms": 1.4827430419921874, "prompt_per_second": 674.4256905474448, "predicted_n": 512, "predicted_ms": 17146.279, "predicted_per_token_ms": 33.55436203522505, "predicted_per_second": 29.802384529028135, "draft_n": 406, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7586206896551724, "mean_accepted_span_including_target_bonus": 2.5172413793103448, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8177339901477833, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6995073891625616 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 406.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 203.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 142.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0099", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.384899233607722, "decode_tps": 29.81937917686492, "first_frame_s": 0.1429872380103916, "first_answer_s": null, "wall_s": 17.279761346057057, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:53:04.222181+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 23, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 140.92, "prompt_per_token_ms": 35.23, "prompt_per_second": 28.384899233607722, "predicted_n": 512, "predicted_ms": 17136.507, "predicted_per_token_ms": 33.53523874755382, "predicted_per_second": 29.81937917686492, "draft_n": 406, "draft_n_accepted": 308 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7586206896551724, "mean_accepted_span_including_target_bonus": 2.5172413793103448, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8177339901477833, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6995073891625616 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 406.0, "llamacpp:spec_decode_num_accepted_tokens_total": 308.0, "llamacpp:spec_decode_num_drafts_total": 203.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 166.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 142.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0100", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 234, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 668.8757857167586, "decode_tps": 31.325271400538096, "first_frame_s": 12.452425417024642, "first_answer_s": 18.37180999503471, "wall_s": 19.890684521989897, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:53:24.172390+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 24, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 12247.416, "prompt_per_token_ms": 1.4950458984375, "prompt_per_second": 668.8757857167586, "predicted_n": 234, "predicted_ms": 7438.084, "predicted_per_token_ms": 31.92310729613734, "predicted_per_second": 31.325271400538096, "draft_n": 176, "draft_n_accepted": 145 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8238636363636364, "mean_accepted_span_including_target_bonus": 2.6477272727272725, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.875, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7727272727272727 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 176.0, "llamacpp:spec_decode_num_accepted_tokens_total": 145.0, "llamacpp:spec_decode_num_drafts_total": 88.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 77.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 68.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0101", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 196, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 655.118942433942, "decode_tps": 28.31100055141118, "first_frame_s": 25.162763239000924, "first_answer_s": 30.25735165493097, "wall_s": 32.05092449299991, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:53:56.301278+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 25, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 25009.199, "prompt_per_token_ms": 1.5264403686523438, "prompt_per_second": 655.118942433942, "predicted_n": 196, "predicted_ms": 6887.782, "predicted_per_token_ms": 35.32195897435898, "predicted_per_second": 28.31100055141118, "draft_n": 142, "draft_n_accepted": 126 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8873239436619719, "mean_accepted_span_including_target_bonus": 2.774647887323944, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9154929577464789, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8591549295774648 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 142.0, "llamacpp:spec_decode_num_accepted_tokens_total": 126.0, "llamacpp:spec_decode_num_drafts_total": 71.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 61.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0102", "run": "guided-speed175-followup-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "hybrid-mtp2-b8192-cache72", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 250, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 619.2538086636182, "decode_tps": 21.92285058335913, "first_frame_s": 46.57119872991461, "first_answer_s": 55.588035363005474, "wall_s": 57.929866853985004, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 72, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T10:54:54.328879+00:00", "configuration_ids": [ "C020" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, "source_row_zero_based": 26, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 46300.886, "prompt_per_token_ms": 1.6148467494419643, "prompt_per_second": 619.2538086636182, "predicted_n": 250, "predicted_ms": 11358.012, "predicted_per_token_ms": 45.614506024096386, "predicted_per_second": 21.92285058335913, "draft_n": 180, "draft_n_accepted": 161 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8944444444444445, "mean_accepted_span_including_target_bonus": 2.7888888888888888, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9444444444444444, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8444444444444444 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 180.0, "llamacpp:spec_decode_num_accepted_tokens_total": 161.0, "llamacpp:spec_decode_num_drafts_total": 90.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 85.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 76.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0103", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 728.3932526884397, "decode_tps": 29.17481962996052, "first_frame_s": 22.496896425029263, "first_answer_s": null, "wall_s": 40.01230204303283, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:05:20.513560+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22493.344, "prompt_per_token_ms": 1.372884765625, "prompt_per_second": 728.3932526884397, "predicted_n": 512, "predicted_ms": 17515.104, "predicted_per_token_ms": 34.27613307240704, "predicted_per_second": 29.17481962996052 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0104", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 31.255860473838844, "decode_tps": 29.277874126111318, "first_frame_s": 0.13158258702605963, "first_answer_s": null, "wall_s": 17.585375369992107, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:05:38.141603+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 127.976, "prompt_per_token_ms": 31.994, "prompt_per_second": 31.255860473838844, "predicted_n": 512, "predicted_ms": 17453.453, "predicted_per_token_ms": 34.15548532289628, "predicted_per_second": 29.277874126111318 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0105", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 738.4554364409549, "decode_tps": 29.203512765392635, "first_frame_s": 22.190182188991457, "first_answer_s": null, "wall_s": 39.68843077297788, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:06:17.873803+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22186.85, "prompt_per_token_ms": 1.3541778564453124, "prompt_per_second": 738.4554364409549, "predicted_n": 512, "predicted_ms": 17497.895, "predicted_per_token_ms": 34.242455968688844, "predicted_per_second": 29.203512765392635 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0106", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 30.377362789249453, "decode_tps": 29.15619739510202, "first_frame_s": 0.13534292893018574, "first_answer_s": null, "wall_s": 17.661939516896382, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:06:35.580460+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 131.677, "prompt_per_token_ms": 32.91925, "prompt_per_second": 30.377362789249453, "predicted_n": 512, "predicted_ms": 17526.291, "predicted_per_token_ms": 34.29802544031311, "predicted_per_second": 29.15619739510202 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0107", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 733.1727370004143, "decode_tps": 29.214515440671807, "first_frame_s": 22.350723957060836, "first_answer_s": null, "wall_s": 39.842323561082594, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:07:15.467567+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22346.712, "prompt_per_token_ms": 1.36393505859375, "prompt_per_second": 733.1727370004143, "predicted_n": 512, "predicted_ms": 17491.305, "predicted_per_token_ms": 34.22955968688846, "predicted_per_second": 29.214515440671807 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0108", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 30.177519257029473, "decode_tps": 29.158120607915436, "first_frame_s": 0.13607460900675505, "first_answer_s": null, "wall_s": 17.661557474988513, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:07:33.171740+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 132.549, "prompt_per_token_ms": 33.13725, "prompt_per_second": 30.177519257029473, "predicted_n": 512, "predicted_ms": 17525.135, "predicted_per_token_ms": 34.29576320939334, "predicted_per_second": 29.158120607915436 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0109", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 229, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 741.775008194658, "decode_tps": 31.674470904358937, "first_frame_s": 11.282092108973302, "first_answer_s": 16.754507174016908, "wall_s": 18.480463362997398, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:07:51.712660+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11043.78, "prompt_per_token_ms": 1.34811767578125, "prompt_per_second": 741.775008194658, "predicted_n": 229, "predicted_ms": 7198.226, "predicted_per_token_ms": 31.571166666666667, "predicted_per_second": 31.674470904358937 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0110", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 214, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 729.0009332315414, "decode_tps": 30.198035312987038, "first_frame_s": 22.617225420079194, "first_answer_s": 27.879771708045155, "wall_s": 29.670996692031622, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:08:21.457596+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22474.594, "prompt_per_token_ms": 1.3717403564453126, "prompt_per_second": 729.0009332315414, "predicted_n": 214, "predicted_ms": 7053.439, "predicted_per_token_ms": 33.11473708920188, "predicted_per_second": 30.198035312987038 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0111", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 266, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 683.1259875687844, "decode_tps": 27.801228142858026, "first_frame_s": 42.20585076708812, "first_answer_s": 49.78476845601108, "wall_s": 51.738438492058776, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:09:13.295969+00:00", "configuration_ids": [ "C022" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 41971.76, "prompt_per_token_ms": 1.4638588169642859, "prompt_per_second": 683.1259875687844, "predicted_n": 266, "predicted_ms": 9531.953, "predicted_per_token_ms": 35.96963396226415, "predicted_per_second": 27.801228142858026 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0112", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 721.1452018205218, "decode_tps": 29.0464224357068, "first_frame_s": 22.72307613794692, "first_answer_s": null, "wall_s": 40.31595078203827, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:09:59.232948+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22719.419, "prompt_per_token_ms": 1.3866832885742189, "prompt_per_second": 721.1452018205218, "predicted_n": 512, "predicted_ms": 17592.528, "predicted_per_token_ms": 34.42764774951076, "predicted_per_second": 29.0464224357068 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0113", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 30.283300274063865, "decode_tps": 29.11362403657264, "first_frame_s": 0.13574673200491816, "first_answer_s": null, "wall_s": 17.688089537085034, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:10:16.963398+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 132.086, "prompt_per_token_ms": 33.0215, "prompt_per_second": 30.283300274063865, "predicted_n": 512, "predicted_ms": 17551.92, "predicted_per_token_ms": 34.34818003913894, "predicted_per_second": 29.11362403657264 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0114", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 735.3114095880228, "decode_tps": 29.065347717969203, "first_frame_s": 22.285230330075137, "first_answer_s": null, "wall_s": 39.86661887704395, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:10:56.871932+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22281.716, "prompt_per_token_ms": 1.359968017578125, "prompt_per_second": 735.3114095880228, "predicted_n": 512, "predicted_ms": 17581.073, "predicted_per_token_ms": 34.40523091976517, "predicted_per_second": 29.065347717969203 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0115", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 31.516163852535872, "decode_tps": 28.998692732121768, "first_frame_s": 0.13074297399725765, "first_answer_s": null, "wall_s": 17.752543167909607, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:11:14.668573+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 126.919, "prompt_per_token_ms": 31.72975, "prompt_per_second": 31.516163852535872, "predicted_n": 512, "predicted_ms": 17621.484, "predicted_per_token_ms": 34.48431311154599, "predicted_per_second": 28.998692732121768 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0116", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 732.2747410445177, "decode_tps": 29.014741703217574, "first_frame_s": 22.377819834975526, "first_answer_s": null, "wall_s": 39.98992324399296, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:11:54.701876+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22374.116, "prompt_per_token_ms": 1.365607666015625, "prompt_per_second": 732.2747410445177, "predicted_n": 512, "predicted_ms": 17611.737, "predicted_per_token_ms": 34.46523874755382, "predicted_per_second": 29.014741703217574 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0117", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 31.23218789284236, "decode_tps": 29.047332201491624, "first_frame_s": 0.13176289503462613, "first_answer_s": null, "wall_s": 17.724093134049326, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:12:12.468751+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 128.073, "prompt_per_token_ms": 32.01825, "prompt_per_second": 31.23218789284236, "predicted_n": 512, "predicted_ms": 17591.977, "predicted_per_token_ms": 34.42656947162426, "predicted_per_second": 29.047332201491624 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0118", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 229, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 740.7409416795813, "decode_tps": 31.79955790246211, "first_frame_s": 11.299527965020388, "first_answer_s": 16.751020528026856, "wall_s": 18.46961641998496, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:12:30.997480+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11059.197, "prompt_per_token_ms": 1.3499996337890625, "prompt_per_second": 740.7409416795813, "predicted_n": 229, "predicted_ms": 7169.911, "predicted_per_token_ms": 31.44697807017544, "predicted_per_second": 31.79955790246211 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0119", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 214, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 728.5992643566607, "decode_tps": 30.16164944856575, "first_frame_s": 22.630873083951883, "first_answer_s": 27.904579736990854, "wall_s": 29.693167435005307, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:13:00.763707+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 16, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22486.984, "prompt_per_token_ms": 1.37249658203125, "prompt_per_second": 728.5992643566607, "predicted_n": 214, "predicted_ms": 7061.948, "predicted_per_token_ms": 33.15468544600939, "predicted_per_second": 30.16164944856575 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0120", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch8192-cache80-t16", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 266, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 681.8353146080696, "decode_tps": 27.689143494336474, "first_frame_s": 42.28948719101027, "first_answer_s": 49.89667164697312, "wall_s": 51.86055858398322, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:13:52.720340+00:00", "configuration_ids": [ "C023" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 17, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 42051.21, "prompt_per_token_ms": 1.4666298130580357, "prompt_per_second": 681.8353146080696, "predicted_n": 266, "predicted_ms": 9570.538, "predicted_per_token_ms": 36.11523773584906, "predicted_per_second": 27.689143494336474 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0121", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 836.0634414165438, "decode_tps": 26.48522919804229, "first_frame_s": 19.60031588107813, "first_answer_s": null, "wall_s": 38.89446728199255, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:14:36.854558+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 18, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 19596.599, "prompt_per_token_ms": 1.1960814819335936, "prompt_per_second": 836.0634414165438, "predicted_n": 512, "predicted_ms": 19293.773, "predicted_per_token_ms": 37.75689432485323, "predicted_per_second": 26.48522919804229 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0122", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 29.959853795913475, "decode_tps": 26.623020889693425, "first_frame_s": 0.13713911699596792, "first_answer_s": null, "wall_s": 19.33140889101196, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:14:56.231187+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 19, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 133.512, "prompt_per_token_ms": 33.378, "prompt_per_second": 29.959853795913475, "predicted_n": 512, "predicted_ms": 19193.915, "predicted_per_token_ms": 37.56147749510763, "predicted_per_second": 26.623020889693425 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0123", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 860.5145061611031, "decode_tps": 26.63710972659535, "first_frame_s": 19.04368635697756, "first_answer_s": null, "wall_s": 38.22778004792053, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:15:34.502307+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 20, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 19039.772, "prompt_per_token_ms": 1.162095458984375, "prompt_per_second": 860.5145061611031, "predicted_n": 512, "predicted_ms": 19183.763, "predicted_per_token_ms": 37.541610567514674, "predicted_per_second": 26.63710972659535 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0124", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 30.25100773669523, "decode_tps": 26.56770240845841, "first_frame_s": 0.1359894439810887, "first_answer_s": null, "wall_s": 19.370243423967622, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:15:53.918854+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 21, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 132.227, "prompt_per_token_ms": 33.05675, "prompt_per_second": 30.25100773669523, "predicted_n": 512, "predicted_ms": 19233.88, "predicted_per_token_ms": 37.639686888454015, "predicted_per_second": 26.56770240845841 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0125", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 859.3896130722056, "decode_tps": 26.661221029661572, "first_frame_s": 19.06826986197848, "first_answer_s": null, "wall_s": 38.23502183298115, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:16:32.195753+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 22, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 19064.694, "prompt_per_token_ms": 1.1636165771484375, "prompt_per_second": 859.3896130722056, "predicted_n": 512, "predicted_ms": 19166.414, "predicted_per_token_ms": 37.50765949119374, "predicted_per_second": 26.661221029661572 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0126", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 29.120771118019203, "decode_tps": 26.629887196631646, "first_frame_s": 0.14107738505117595, "first_answer_s": null, "wall_s": 19.330419238074683, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:16:51.570274+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 23, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 137.359, "prompt_per_token_ms": 34.33975, "prompt_per_second": 29.120771118019203, "predicted_n": 512, "predicted_ms": 19188.966, "predicted_per_token_ms": 37.551792563600785, "predicted_per_second": 26.629887196631646 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input16384/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0127", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 267, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 737.4627949029726, "decode_tps": 28.16991798000084, "first_frame_s": 11.341006714035757, "first_answer_s": 18.96795085608028, "wall_s": 20.783831560984254, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:17:12.412860+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 24, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11108.357, "prompt_per_token_ms": 1.3560006103515625, "prompt_per_second": 737.4627949029726, "predicted_n": 267, "predicted_ms": 9442.697, "predicted_per_token_ms": 35.498860902255636, "predicted_per_second": 28.16991798000084 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0128", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 198, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 860.5545513534773, "decode_tps": 27.011681249546665, "first_frame_s": 19.183133386075497, "first_answer_s": 24.60555438697338, "wall_s": 26.47660952003207, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:17:38.962081+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 25, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 19038.886, "prompt_per_token_ms": 1.1620413818359374, "prompt_per_second": 860.5545513534773, "predicted_n": 198, "predicted_ms": 7293.141, "predicted_per_token_ms": 37.02102030456852, "predicted_per_second": 27.011681249546665 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0129", "run": "guided-speed175-native-profile-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "batch16384-cache16-t16", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 233, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 803.7063330164142, "decode_tps": 25.14148315461026, "first_frame_s": 35.86995446495712, "first_answer_s": 43.08858014794532, "wall_s": 45.09842677600682, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": 16, "cpu_moe": "99", "batch": 16384, "ubatch": "16384", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 16, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:18:24.159678+00:00", "configuration_ids": [ "C021" ], "missing_configuration_match": false, "source": { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, "source_row_zero_based": 26, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 35674.722, "prompt_per_token_ms": 1.2442355608258928, "prompt_per_second": 803.7063330164142, "predicted_n": 233, "predicted_ms": 9227.777, "predicted_per_token_ms": 39.77490086206897, "predicted_per_second": 25.14148315461026 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0130", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 135.97496799102444, "decode_tps": 18.924018805364096, "first_frame_s": 60.448522896971554, "first_answer_s": null, "wall_s": 87.45116588496603, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:01:45.667843+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 60246.383, "prompt_per_token_ms": 7.354294799804688, "prompt_per_second": 135.97496799102444, "predicted_n": 512, "predicted_ms": 27002.721, "predicted_per_token_ms": 52.84289823874756, "predicted_per_second": 18.924018805364096 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0131", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 25.84313218762114, "decode_tps": 19.1032707865082, "first_frame_s": 0.15681684704031795, "first_answer_s": null, "wall_s": 26.906548556988128, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:02:12.611380+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 154.78, "prompt_per_token_ms": 38.695, "prompt_per_second": 25.84313218762114, "predicted_n": 512, "predicted_ms": 26749.346, "predicted_per_token_ms": 52.34705675146771, "predicted_per_second": 19.1032707865082 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0132", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 137.92579306110355, "decode_tps": 19.765210954367273, "first_frame_s": 59.39643986092415, "first_answer_s": null, "wall_s": 85.25020701193716, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:03:37.901647+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59394.257, "prompt_per_token_ms": 7.250275512695312, "prompt_per_second": 137.92579306110355, "predicted_n": 512, "predicted_ms": 25853.506, "predicted_per_token_ms": 50.59394520547946, "predicted_per_second": 19.765210954367273 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0133", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 24.87979946881628, "decode_tps": 20.100038402480415, "first_frame_s": 0.16326227993704379, "first_answer_s": null, "wall_s": 25.586398171028122, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:04:03.526446+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 160.773, "prompt_per_token_ms": 40.19325, "prompt_per_second": 24.87979946881628, "predicted_n": 512, "predicted_ms": 25422.837, "predicted_per_token_ms": 49.751148727984344, "predicted_per_second": 20.100038402480415 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0134", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 137.87256075005502, "decode_tps": 19.12872154705613, "first_frame_s": 59.419240541988984, "first_answer_s": null, "wall_s": 86.13324229698628, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:05:29.699532+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59417.189, "prompt_per_token_ms": 7.253074829101562, "prompt_per_second": 137.87256075005502, "predicted_n": 512, "predicted_ms": 26713.756, "predicted_per_token_ms": 52.27740900195695, "predicted_per_second": 19.12872154705613 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0135", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 24.423304717361308, "decode_tps": 19.372168123578813, "first_frame_s": 0.16597582306712866, "first_answer_s": null, "wall_s": 26.544317853986286, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:05:56.282070+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 163.778, "prompt_per_token_ms": 40.9445, "prompt_per_second": 24.423304717361308, "predicted_n": 512, "predicted_ms": 26378.049, "predicted_per_token_ms": 51.620448140900194, "predicted_per_second": 19.372168123578813 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0136", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "context-quality", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 196, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "unknown", "prefill_tps": 136.68407992701685, "decode_tps": 20.026307378965107, "first_frame_s": 60.12144565396011, "first_answer_s": null, "wall_s": 69.85887019801885, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:07:06.196367+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59933.827, "prompt_per_token_ms": 7.316140991210937, "prompt_per_second": 136.68407992701685, "predicted_n": 196, "predicted_ms": 9737.192, "predicted_per_token_ms": 49.93431794871795, "predicted_per_second": 20.026307378965107 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "truncated": false } }, { "id": "M0137", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "context-quality", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 192, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "unknown", "prefill_tps": 137.0880670510588, "decode_tps": 19.79474815276654, "first_frame_s": 59.759238964063115, "first_answer_s": null, "wall_s": 69.40863966301549, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:08:15.643780+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59757.207, "prompt_per_token_ms": 7.294580932617188, "prompt_per_second": 137.0880670510588, "predicted_n": 192, "predicted_ms": 9649.024, "predicted_per_token_ms": 50.5184502617801, "predicted_per_second": 19.79474815276654 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "truncated": false } }, { "id": "M0138", "run": "http-baseline6-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "context-quality", "case": "http-baseline6-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 222, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "unknown", "prefill_tps": 137.1042491371654, "decode_tps": 20.019300054613737, "first_frame_s": 59.75245083705522, "first_answer_s": null, "wall_s": 70.79209305497352, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:09:26.475754+00:00", "configuration_ids": [ "C025" ], "missing_configuration_match": false, "source": { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59750.154, "prompt_per_token_ms": 7.293719970703125, "prompt_per_second": 137.1042491371654, "predicted_n": 222, "predicted_ms": 11039.347, "predicted_per_token_ms": 49.9517963800905, "predicted_per_second": 20.019300054613737 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "truncated": false } }, { "id": "M0139", "run": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-batch1024-cache32-vision-20261005", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 219.6513410341792, "decode_tps": 19.711903015276956, "first_frame_s": 37.432222625939175, "first_answer_s": null, "wall_s": 63.35588111798279, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:16:34.530850+00:00", "configuration_ids": [ "C026" ], "missing_configuration_match": false, "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 37295.47, "prompt_per_token_ms": 4.552669677734375, "prompt_per_second": 219.6513410341792, "predicted_n": 512, "predicted_ms": 25923.423, "predicted_per_token_ms": 50.73076908023483, "predicted_per_second": 19.711903015276956 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0140", "run": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-batch1024-cache32-vision-20261005", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 26.910294533173666, "decode_tps": 19.92263778531381, "first_frame_s": 0.15107160597108305, "first_answer_s": null, "wall_s": 25.80050667701289, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:17:00.373979+00:00", "configuration_ids": [ "C026" ], "missing_configuration_match": false, "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 148.642, "prompt_per_token_ms": 37.1605, "prompt_per_second": 26.910294533173666, "predicted_n": 512, "predicted_ms": 25649.214, "predicted_per_token_ms": 50.19415655577299, "predicted_per_second": 19.92263778531381 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0141", "run": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-batch1024-cache32-vision-20261005", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 225.27059213689589, "decode_tps": 19.79701972333297, "first_frame_s": 36.36733056406956, "first_answer_s": null, "wall_s": 62.17954571999144, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:18:02.591532+00:00", "configuration_ids": [ "C026" ], "missing_configuration_match": false, "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 36365.155, "prompt_per_token_ms": 4.439105834960937, "prompt_per_second": 225.27059213689589, "predicted_n": 512, "predicted_ms": 25811.966, "predicted_per_token_ms": 50.51265362035225, "predicted_per_second": 19.79701972333297 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0142", "run": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-batch1024-cache32-vision-20261005", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.05226136291912, "decode_tps": 20.134307255911814, "first_frame_s": 0.14459146198350936, "first_answer_s": null, "wall_s": 25.524391956045292, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:18:28.154249+00:00", "configuration_ids": [ "C026" ], "missing_configuration_match": false, "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 142.591, "prompt_per_token_ms": 35.64775, "prompt_per_second": 28.05226136291912, "predicted_n": 512, "predicted_ms": 25379.567, "predicted_per_token_ms": 49.666471624266144, "predicted_per_second": 20.134307255911814 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0143", "run": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-batch1024-cache32-vision-20261005", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 224.31239092940052, "decode_tps": 20.311497281300124, "first_frame_s": 36.52266518503893, "first_answer_s": null, "wall_s": 61.68127716402523, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:19:29.874788+00:00", "configuration_ids": [ "C026" ], "missing_configuration_match": false, "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 36520.497, "prompt_per_token_ms": 4.458068481445313, "prompt_per_second": 224.31239092940052, "predicted_n": 512, "predicted_ms": 25158.165, "predicted_per_token_ms": 49.23319960861057, "predicted_per_second": 20.311497281300124 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0144", "run": "http-batch1024-cache32-vision-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-batch1024-cache32-vision-20261005", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.73232961728537, "decode_tps": 20.484727172697955, "first_frame_s": 0.14132876694202423, "first_answer_s": null, "wall_s": 25.086986107984558, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:19:54.999150+00:00", "configuration_ids": [ "C026" ], "missing_configuration_match": false, "source": { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 139.216, "prompt_per_token_ms": 34.804, "prompt_per_second": 28.73232961728537, "predicted_n": 512, "predicted_ms": 24945.414, "predicted_per_token_ms": 48.816857142857145, "predicted_per_second": 20.484727172697955 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0145", "run": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-context32-cache32-u1024-20261005", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 223.9516999169695, "decode_tps": 19.89508167106227, "first_frame_s": 36.71900167898275, "first_answer_s": null, "wall_s": 62.4039983630646, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:32:40.632621+00:00", "configuration_ids": [ "C027" ], "missing_configuration_match": false, "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 36579.316, "prompt_per_token_ms": 4.46524853515625, "prompt_per_second": 223.9516999169695, "predicted_n": 512, "predicted_ms": 25684.74, "predicted_per_token_ms": 50.26367906066537, "predicted_per_second": 19.89508167106227 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0146", "run": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-context32-cache32-u1024-20261005", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.604528096797726, "decode_tps": 19.837966902662124, "first_frame_s": 0.14218002697452903, "first_answer_s": null, "wall_s": 25.901084744022228, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:33:06.571939+00:00", "configuration_ids": [ "C027" ], "missing_configuration_match": false, "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 139.838, "prompt_per_token_ms": 34.9595, "prompt_per_second": 28.604528096797726, "predicted_n": 512, "predicted_ms": 25758.688, "predicted_per_token_ms": 50.40839138943248, "predicted_per_second": 19.837966902662124 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0147", "run": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-context32-cache32-u1024-20261005", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 224.8140358164545, "decode_tps": 20.086194538371096, "first_frame_s": 36.441216803039424, "first_answer_s": null, "wall_s": 61.88182112306822, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:34:08.493100+00:00", "configuration_ids": [ "C027" ], "missing_configuration_match": false, "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 36439.006, "prompt_per_token_ms": 4.448120849609375, "prompt_per_second": 224.8140358164545, "predicted_n": 512, "predicted_ms": 25440.359, "predicted_per_token_ms": 49.785438356164384, "predicted_per_second": 20.086194538371096 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0148", "run": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-context32-cache32-u1024-20261005", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 26.619992945701867, "decode_tps": 20.098470706970588, "first_frame_s": 0.1523604680551216, "first_answer_s": null, "wall_s": 25.57738233206328, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:34:34.108007+00:00", "configuration_ids": [ "C027" ], "missing_configuration_match": false, "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 150.263, "prompt_per_token_ms": 37.56575, "prompt_per_second": 26.619992945701867, "predicted_n": 512, "predicted_ms": 25424.82, "predicted_per_token_ms": 49.75502935420744, "predicted_per_second": 20.098470706970588 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0149", "run": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-context32-cache32-u1024-20261005", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 224.38426418792906, "decode_tps": 20.470197635551965, "first_frame_s": 36.51112818601541, "first_answer_s": null, "wall_s": 61.47448069206439, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:35:35.622426+00:00", "configuration_ids": [ "C027" ], "missing_configuration_match": false, "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 36508.799, "prompt_per_token_ms": 4.456640502929687, "prompt_per_second": 224.38426418792906, "predicted_n": 512, "predicted_ms": 24963.12, "predicted_per_token_ms": 48.851506849315065, "predicted_per_second": 20.470197635551965 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0150", "run": "http-context32-cache32-u1024-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-context32-cache32-u1024-20261005", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.382885120272473, "decode_tps": 20.47492284740545, "first_frame_s": 0.14333584799896926, "first_answer_s": null, "wall_s": 25.100918764946982, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T06:36:00.762638+00:00", "configuration_ids": [ "C027" ], "missing_configuration_match": false, "source": { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 140.93, "prompt_per_token_ms": 35.2325, "prompt_per_second": 28.382885120272473, "predicted_n": 512, "predicted_ms": 24957.359, "predicted_per_token_ms": 48.84023287671233, "predicted_per_second": 20.47492284740545 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0151", "run": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-threads14-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 138.10900516232985, "decode_tps": 19.628073738102113, "first_frame_s": 59.51203510002233, "first_answer_s": null, "wall_s": 85.5463996110484, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:16:20.178279+00:00", "configuration_ids": [ "C028" ], "missing_configuration_match": false, "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59315.466, "prompt_per_token_ms": 7.240657470703125, "prompt_per_second": 138.10900516232985, "predicted_n": 512, "predicted_ms": 26034.139, "predicted_per_token_ms": 50.94743444227006, "predicted_per_second": 19.628073738102113 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0152", "run": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-threads14-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.50606822927431, "decode_tps": 19.61028793313744, "first_frame_s": 0.1429221259895712, "first_answer_s": null, "wall_s": 26.200834693969227, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:16:46.417674+00:00", "configuration_ids": [ "C028" ], "missing_configuration_match": false, "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 140.321, "prompt_per_token_ms": 35.08025, "prompt_per_second": 28.50606822927431, "predicted_n": 512, "predicted_ms": 26057.751, "predicted_per_token_ms": 50.99364187866928, "predicted_per_second": 19.61028793313744 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0153", "run": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-threads14-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 137.94958353606745, "decode_tps": 20.44456977023504, "first_frame_s": 59.38631085399538, "first_answer_s": null, "wall_s": 84.3809008699609, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:18:10.835799+00:00", "configuration_ids": [ "C028" ], "missing_configuration_match": false, "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59384.014, "prompt_per_token_ms": 7.249025146484375, "prompt_per_second": 137.94958353606745, "predicted_n": 512, "predicted_ms": 24994.412, "predicted_per_token_ms": 48.91274363992172, "predicted_per_second": 20.44456977023504 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0154", "run": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-threads14-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.969972623375867, "decode_tps": 20.4785775443781, "first_frame_s": 0.14040725002996624, "first_answer_s": null, "wall_s": 25.093469239072874, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:18:35.969033+00:00", "configuration_ids": [ "C028" ], "missing_configuration_match": false, "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 138.074, "prompt_per_token_ms": 34.5185, "prompt_per_second": 28.969972623375867, "predicted_n": 512, "predicted_ms": 24952.905, "predicted_per_token_ms": 48.83151663405088, "predicted_per_second": 20.4785775443781 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0155", "run": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-threads14-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 138.02165302293199, "decode_tps": 19.75518798878385, "first_frame_s": 59.35556529497262, "first_answer_s": null, "wall_s": 85.22237887396477, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:20:01.231224+00:00", "configuration_ids": [ "C028" ], "missing_configuration_match": false, "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 59353.006, "prompt_per_token_ms": 7.245239990234375, "prompt_per_second": 138.02165302293199, "predicted_n": 512, "predicted_ms": 25866.623, "predicted_per_token_ms": 50.619614481409, "predicted_per_second": 19.75518798878385 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0156", "run": "http-threads14-cache32-vision-20261004", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "http-threads14-cache32-vision-20261004", "context_tokens": 131072, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.6582220439044, "decode_tps": 19.72639528345363, "first_frame_s": 0.14193322404753417, "first_answer_s": null, "wall_s": 26.046515246038325, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-04T22:20:27.317398+00:00", "configuration_ids": [ "C028" ], "missing_configuration_match": false, "source": { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 139.576, "prompt_per_token_ms": 34.894, "prompt_per_second": 28.6582220439044, "predicted_n": 512, "predicted_ms": 25904.378, "predicted_per_token_ms": 50.69349902152642, "predicted_per_second": 19.72639528345363 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0157", "run": "ik178/results/control32-noMTP", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ik-ctx32768-cpu43-f16-u8192-MTP0", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 298, "fresh_input_tokens": 8192, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 1093.2393581649378, "decode_tps": 31.055434575995353, "first_frame_s": 7.496807857998647, "first_answer_s": 15.475937849027105, "wall_s": 17.092680150992237, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 0, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:02:42.345015+00:00", "configuration_ids": [ "C029" ], "missing_configuration_match": false, "source": { "path": "ik178/results/control32-noMTP/requests.json", "bytes": 36758, "sha256": "ca40d5e5b17d4f6f759460c4acdac5fbd606fb28cbbd3acb8284d11c7053356e" }, "source_row_zero_based": 0, "details": { "timings": { "prompt_n": 8192, "prompt_ms": 7493.327, "prompt_per_token_ms": 0.9147127685546875, "prompt_per_second": 1093.2393581649378, "predicted_n": 298, "predicted_ms": 9595.744, "predicted_per_token_ms": 32.20048322147651, "predicted_per_second": 31.055434575995353, "n_ctx": 32768, "n_past": 8489 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 0, "native_accepted_draft_tokens": 0 }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "reference_receipt_sha256": "8432830f67f966a456595d818b0ce8be758161f37e6eb2a9ebf888b4578781ad", "reference_payload_sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0158", "run": "ik178/results/control32-noMTP", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ik-ctx32768-cpu43-f16-u8192-MTP0", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 207, "fresh_input_tokens": 28672, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 1089.7492421317004, "decode_tps": 30.913369091290317, "first_frame_s": 26.31850827904418, "first_answer_s": 31.408315518056042, "wall_s": 33.015457367990166, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 0, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:03:15.426663+00:00", "configuration_ids": [ "C029" ], "missing_configuration_match": false, "source": { "path": "ik178/results/control32-noMTP/requests.json", "bytes": 36758, "sha256": "ca40d5e5b17d4f6f759460c4acdac5fbd606fb28cbbd3acb8284d11c7053356e" }, "source_row_zero_based": 1, "details": { "timings": { "prompt_n": 28672, "prompt_ms": 26310.64, "prompt_per_token_ms": 0.9176422991071428, "prompt_per_second": 1089.7492421317004, "predicted_n": 207, "predicted_ms": 6696.132, "predicted_per_token_ms": 32.34846376811594, "predicted_per_second": 30.913369091290317, "n_ctx": 32768, "n_past": 28878 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 0, "native_accepted_draft_tokens": 0 }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "reference_receipt_sha256": "eb3dcaf20443e43ca1c94f21c333fb6fa04993276ac8a8ed09f36d6ae4d53e52", "reference_payload_sha256": "3cba61a6792cf8982286162eb252d55c2bb51c905e30dcbfa7622c3169c3c917", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 4064, "truncated": false } }, { "id": "M0159", "run": "ik178/results/control32-noMTP", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ik-ctx32768-cpu43-f16-u8192-MTP0", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 1223.08361296391, "decode_tps": 30.944034812039163, "first_frame_s": 6.701010520919226, "first_answer_s": null, "wall_s": 23.246994526009075, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 0, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:03:38.744599+00:00", "configuration_ids": [ "C029" ], "missing_configuration_match": false, "source": { "path": "ik178/results/control32-noMTP/requests.json", "bytes": 36758, "sha256": "ca40d5e5b17d4f6f759460c4acdac5fbd606fb28cbbd3acb8284d11c7053356e" }, "source_row_zero_based": 2, "details": { "timings": { "prompt_n": 8192, "prompt_ms": 6697.825, "prompt_per_token_ms": 0.8176055908203125, "prompt_per_second": 1223.08361296391, "predicted_n": 512, "predicted_ms": 16546.0, "predicted_per_token_ms": 32.31640625, "predicted_per_second": 30.944034812039163, "n_ctx": 32768, "n_past": 8703 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 0, "native_accepted_draft_tokens": 0 }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "reference_receipt_sha256": "e6a49f65bba6f20df86fe57383e620b42e00395b994b680e307774094889c328", "reference_payload_sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0160", "run": "ik178/results/control32-noMTP", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ik-ctx32768-cpu43-f16-u8192-MTP0", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 512, "fresh_input_tokens": 28672, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 1089.8358138323854, "decode_tps": 30.732145343839882, "first_frame_s": 26.316806601011194, "first_answer_s": null, "wall_s": 42.97757345298305, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 0, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:04:21.786913+00:00", "configuration_ids": [ "C029" ], "missing_configuration_match": false, "source": { "path": "ik178/results/control32-noMTP/requests.json", "bytes": 36758, "sha256": "ca40d5e5b17d4f6f759460c4acdac5fbd606fb28cbbd3acb8284d11c7053356e" }, "source_row_zero_based": 3, "details": { "timings": { "prompt_n": 28672, "prompt_ms": 26308.55, "prompt_per_token_ms": 0.9175694056919642, "prompt_per_second": 1089.8358138323854, "predicted_n": 512, "predicted_ms": 16660.08, "predicted_per_token_ms": 32.53921875, "predicted_per_second": 30.732145343839882, "n_ctx": 32768, "n_past": 29183 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 0, "native_accepted_draft_tokens": 0 }, "input_token_sha256": "eab266f2d9bc6ee360c81ab6177085a436334063ad5c0251766de3421d1528d4", "reference_receipt_sha256": "220d8092d30850024ca0ddd66dcca8405bc6f7cba5e0ce9adfb9d0ae85df2691", "reference_payload_sha256": "19766cf4c14b17729e8087d3260c4a66034d819410f62423c9f721a6af965b3b", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0161", "run": "ik178/results/mtp32-n1-mainU4096-draftU512", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ik-ctx32768-cpu43-f16-u4096-MTP1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 229, "fresh_input_tokens": 8192, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 690.5125580649667, "decode_tps": 37.54855487253987, "first_frame_s": 11.86687953909859, "first_answer_s": 16.75581375102047, "wall_s": 17.96754529606551, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 1, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:16:02.403273+00:00", "configuration_ids": [ "C032" ], "missing_configuration_match": false, "source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.json", "bytes": 40991, "sha256": "b0c4b235b06560ccf360509232909a39c3a61c50623d801034a31ca09e12f274" }, "source_row_zero_based": 0, "details": { "timings": { "prompt_n": 8192, "prompt_ms": 11863.651, "prompt_per_token_ms": 1.4481995849609375, "prompt_per_second": 690.5125580649667, "predicted_n": 229, "predicted_ms": 6098.77, "predicted_per_token_ms": 26.63218340611354, "predicted_per_second": 37.54855487253987, "n_ctx": 32768, "n_past": 8421, "draft_n": 124, "draft_n_accepted": 105, "draft_by_depth": [ { "depth": 1, "draft_n": 124, "draft_n_accepted": 105 } ] }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 124, "native_accepted_draft_tokens": 105 }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "reference_receipt_sha256": "8432830f67f966a456595d818b0ce8be758161f37e6eb2a9ebf888b4578781ad", "reference_payload_sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0162", "run": "ik178/results/mtp32-n1-mainU4096-draftU512", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "ik-ctx32768-cpu43-f16-u4096-MTP1", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 232, "fresh_input_tokens": 28672, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 786.1079330688536, "decode_tps": 33.96443747718014, "first_frame_s": 36.481240329914726, "first_answer_s": 42.00274185498711, "wall_s": 43.314450152916834, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 1, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:16:45.781216+00:00", "configuration_ids": [ "C032" ], "missing_configuration_match": false, "source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.json", "bytes": 40991, "sha256": "b0c4b235b06560ccf360509232909a39c3a61c50623d801034a31ca09e12f274" }, "source_row_zero_based": 1, "details": { "timings": { "prompt_n": 28672, "prompt_ms": 36473.363, "prompt_per_token_ms": 1.2720899483816963, "prompt_per_second": 786.1079330688536, "predicted_n": 232, "predicted_ms": 6830.674, "predicted_per_token_ms": 29.442560344827587, "predicted_per_second": 33.96443747718014, "n_ctx": 32768, "n_past": 28903, "draft_n": 130, "draft_n_accepted": 101, "draft_by_depth": [ { "depth": 1, "draft_n": 130, "draft_n_accepted": 101 } ] }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 130, "native_accepted_draft_tokens": 101 }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "reference_receipt_sha256": "eb3dcaf20443e43ca1c94f21c333fb6fa04993276ac8a8ed09f36d6ae4d53e52", "reference_payload_sha256": "3cba61a6792cf8982286162eb252d55c2bb51c905e30dcbfa7622c3169c3c917", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 4064, "truncated": false } }, { "id": "M0163", "run": "ik178/results/mtp32-n1-mainU4096-draftU512", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ik-ctx32768-cpu43-f16-u4096-MTP1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 911.1286940362667, "decode_tps": 33.87865394020325, "first_frame_s": 8.994098813040182, "first_answer_s": null, "wall_s": 24.108653538976796, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 1, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:17:09.963317+00:00", "configuration_ids": [ "C032" ], "missing_configuration_match": false, "source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.json", "bytes": 40991, "sha256": "b0c4b235b06560ccf360509232909a39c3a61c50623d801034a31ca09e12f274" }, "source_row_zero_based": 2, "details": { "timings": { "prompt_n": 8192, "prompt_ms": 8991.046, "prompt_per_token_ms": 1.097539794921875, "prompt_per_second": 911.1286940362667, "predicted_n": 512, "predicted_ms": 15112.761, "predicted_per_token_ms": 29.517111328125, "predicted_per_second": 33.87865394020325, "n_ctx": 32768, "n_past": 8703, "draft_n": 307, "draft_n_accepted": 204, "draft_by_depth": [ { "depth": 1, "draft_n": 307, "draft_n_accepted": 204 } ] }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 307, "native_accepted_draft_tokens": 204 }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "reference_receipt_sha256": "e6a49f65bba6f20df86fe57383e620b42e00395b994b680e307774094889c328", "reference_payload_sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0164", "run": "ik178/results/mtp32-n1-mainU4096-draftU512", "engine": "ik_llama.cpp", "commit": "3d27f5bb6348c4d79edded63b03e9a93efb2f80e", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "ik-ctx32768-cpu43-f16-u4096-MTP1", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 512, "fresh_input_tokens": 28672, "cached_input_tokens": null, "cache_kind": "fresh-prompt", "prefill_tps": 851.6968839544281, "decode_tps": 34.26291568043674, "first_frame_s": 33.672981166047975, "first_answer_s": null, "wall_s": 48.65918868605513, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 1, "load_mode": null, "thinking": "native ON", "time_utc": "2026-10-05T19:17:58.687059+00:00", "configuration_ids": [ "C032" ], "missing_configuration_match": false, "source": { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.json", "bytes": 40991, "sha256": "b0c4b235b06560ccf360509232909a39c3a61c50623d801034a31ca09e12f274" }, "source_row_zero_based": 3, "details": { "timings": { "prompt_n": 28672, "prompt_ms": 33664.559, "prompt_per_token_ms": 1.1741266392299108, "prompt_per_second": 851.6968839544281, "predicted_n": 512, "predicted_ms": 14943.27, "predicted_per_token_ms": 29.18607421875, "predicted_per_second": 34.26291568043674, "n_ctx": 32768, "n_past": 29183, "draft_n": 286, "draft_n_accepted": 224, "draft_by_depth": [ { "depth": 1, "draft_n": 286, "draft_n_accepted": 224 } ] }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "fresh_verified_by": "cache_prompt:false and prompt_n equals exact input", "native_draft_tokens": 286, "native_accepted_draft_tokens": 224 }, "input_token_sha256": "eab266f2d9bc6ee360c81ab6177085a436334063ad5c0251766de3421d1528d4", "reference_receipt_sha256": "220d8092d30850024ca0ddd66dcca8405bc6f7cba5e0ce9adfb9d0ae85df2691", "reference_payload_sha256": "19766cf4c14b17729e8087d3260c4a66034d819410f62423c9f721a6af965b3b", "chat_template_sha256": "12827f24b742ea4e80cdc12dbcf9622227056b9f797252a3149263d4f9aaadce", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0165", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 153.96036347576046, "decode_tps": 25.60127679729731, "first_frame_s": 53.21118436998222, "first_answer_s": null, "wall_s": 73.17089010495692, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:43:03.128727+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 53208.5, "prompt_per_token_ms": 6.49517822265625, "prompt_per_second": 153.96036347576046, "predicted_n": 512, "predicted_ms": 19959.942, "predicted_per_token_ms": 39.0605518590998, "predicted_per_second": 25.60127679729731 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0166", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 24.918857968739296, "decode_tps": 26.853951343477963, "first_frame_s": 0.16275122400838882, "first_answer_s": null, "wall_s": 19.191825138987042, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:43:22.362114+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 160.521, "prompt_per_token_ms": 40.13025, "prompt_per_second": 24.918857968739296, "predicted_n": 512, "predicted_ms": 19028.857, "predicted_per_token_ms": 37.23846771037182, "predicted_per_second": 26.853951343477963 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0167", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 183.06861980384127, "decode_tps": 26.845875934962795, "first_frame_s": 44.75023569702171, "first_answer_s": null, "wall_s": 63.78504327801056, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:44:26.187273+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 44748.248, "prompt_per_token_ms": 5.4624326171875, "prompt_per_second": 183.06861980384127, "predicted_n": 512, "predicted_ms": 19034.581, "predicted_per_token_ms": 37.24966927592955, "predicted_per_second": 26.845875934962795 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0168", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 24.642529309208296, "decode_tps": 26.76213749349277, "first_frame_s": 0.1642445019679144, "first_answer_s": null, "wall_s": 19.258666991023347, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:44:45.488290+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 162.321, "prompt_per_token_ms": 40.58025, "prompt_per_second": 24.642529309208296, "predicted_n": 512, "predicted_ms": 19094.14, "predicted_per_token_ms": 37.366223091976515, "predicted_per_second": 26.76213749349277 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0169", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 183.26640783567697, "decode_tps": 26.728225623465935, "first_frame_s": 44.70197176199872, "first_answer_s": null, "wall_s": 63.82063777698204, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:45:49.352434+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 44699.954, "prompt_per_token_ms": 5.456537353515625, "prompt_per_second": 183.26640783567697, "predicted_n": 512, "predicted_ms": 19118.366, "predicted_per_token_ms": 37.413632093933465, "predicted_per_second": 26.728225623465935 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0170", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 24.679475314354818, "decode_tps": 26.5562412077924, "first_frame_s": 0.1639714720658958, "first_answer_s": null, "wall_s": 19.406496667070314, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:46:08.802878+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 162.078, "prompt_per_token_ms": 40.5195, "prompt_per_second": 24.679475314354818, "predicted_n": 512, "predicted_ms": 19242.181, "predicted_per_token_ms": 37.65593150684931, "predicted_per_second": 26.5562412077924 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0171", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "author-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 228, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 181.47799114142833, "decode_tps": 27.022445533701628, "first_frame_s": 45.32990919495933, "first_answer_s": 51.696930126985535, "wall_s": 53.73062292591203, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:47:02.592562+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 45140.46, "prompt_per_token_ms": 5.51031005859375, "prompt_per_second": 181.47799114142833, "predicted_n": 228, "predicted_ms": 8400.424, "predicted_per_token_ms": 37.00627312775331, "predicted_per_second": 27.022445533701628 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0172", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "author-control", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 208, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 177.48630620675485, "decode_tps": 26.10203550482093, "first_frame_s": 92.50015258498024, "first_answer_s": 98.34856201498769, "wall_s": 100.43094881495927, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:48:43.099116+00:00", "configuration_ids": [ "C033" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 92311.347, "prompt_per_token_ms": 5.634237487792968, "prompt_per_second": 177.48630620675485, "predicted_n": 208, "predicted_ms": 7930.416, "predicted_per_token_ms": 38.311188405797104, "predicted_per_second": 26.10203550482093 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0173", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 172.31598190766329, "decode_tps": 24.906878251328536, "first_frame_s": 47.54310869600158, "first_answer_s": null, "wall_s": 68.05983872106299, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:49:57.539125+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 47540.57, "prompt_per_token_ms": 5.803292236328125, "prompt_per_second": 172.31598190766329, "predicted_n": 512, "predicted_ms": 20516.421, "predicted_per_token_ms": 40.1495518590998, "predicted_per_second": 24.906878251328536, "draft_n": 277, "draft_n_accepted": 234 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8447653429602888, "mean_accepted_span_including_target_bonus": 1.8447653429602888, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8447653429602888 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 277.0, "llamacpp:spec_decode_num_accepted_tokens_total": 234.0, "llamacpp:spec_decode_num_drafts_total": 277.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 234.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0174", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 22.09554111980202, "decode_tps": 25.200754089414154, "first_frame_s": 0.18310097500216216, "first_answer_s": null, "wall_s": 20.460627270047553, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:50:18.044619+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 181.032, "prompt_per_token_ms": 45.258, "prompt_per_second": 22.09554111980202, "predicted_n": 512, "predicted_ms": 20277.171, "predicted_per_token_ms": 39.68135225048923, "predicted_per_second": 25.200754089414154, "draft_n": 275, "draft_n_accepted": 236 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8581818181818182, "mean_accepted_span_including_target_bonus": 1.8581818181818182, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8581818181818182 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 275.0, "llamacpp:spec_decode_num_accepted_tokens_total": 236.0, "llamacpp:spec_decode_num_drafts_total": 275.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 236.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0175", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 174.34364682131434, "decode_tps": 25.350973311179924, "first_frame_s": 46.99000850494485, "first_answer_s": null, "wall_s": 67.14729790901765, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:51:25.233998+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 46987.66, "prompt_per_token_ms": 5.73579833984375, "prompt_per_second": 174.34364682131434, "predicted_n": 512, "predicted_ms": 20157.017, "predicted_per_token_ms": 39.44621722113503, "predicted_per_second": 25.350973311179924, "draft_n": 276, "draft_n_accepted": 235 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8514492753623188, "mean_accepted_span_including_target_bonus": 1.8514492753623188, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8514492753623188 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 276.0, "llamacpp:spec_decode_num_accepted_tokens_total": 235.0, "llamacpp:spec_decode_num_drafts_total": 276.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 235.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0176", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 22.11435332102301, "decode_tps": 25.138526316618016, "first_frame_s": 0.18298959906678647, "first_answer_s": null, "wall_s": 20.51074460404925, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:51:45.787590+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 180.878, "prompt_per_token_ms": 45.2195, "prompt_per_second": 22.11435332102301, "predicted_n": 512, "predicted_ms": 20327.365, "predicted_per_token_ms": 39.77957925636008, "predicted_per_second": 25.138526316618016, "draft_n": 277, "draft_n_accepted": 234 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8447653429602888, "mean_accepted_span_including_target_bonus": 1.8447653429602888, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8447653429602888 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 277.0, "llamacpp:spec_decode_num_accepted_tokens_total": 234.0, "llamacpp:spec_decode_num_drafts_total": 277.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 234.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0177", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 174.22792168335818, "decode_tps": 25.372645209982224, "first_frame_s": 47.02104788494762, "first_answer_s": null, "wall_s": 67.16116033494473, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:52:52.992420+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 47018.87, "prompt_per_token_ms": 5.739608154296875, "prompt_per_second": 174.22792168335818, "predicted_n": 512, "predicted_ms": 20139.8, "predicted_per_token_ms": 39.41252446183953, "predicted_per_second": 25.372645209982224, "draft_n": 276, "draft_n_accepted": 235 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8514492753623188, "mean_accepted_span_including_target_bonus": 1.8514492753623188, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8514492753623188 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 276.0, "llamacpp:spec_decode_num_accepted_tokens_total": 235.0, "llamacpp:spec_decode_num_drafts_total": 276.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 235.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0178", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 21.56496986295462, "decode_tps": 25.172861448422804, "first_frame_s": 0.1876210740301758, "first_answer_s": null, "wall_s": 20.48755923996214, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:53:13.523384+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 185.486, "prompt_per_token_ms": 46.3715, "prompt_per_second": 21.56496986295462, "predicted_n": 512, "predicted_ms": 20299.639, "predicted_per_token_ms": 39.72532093933464, "predicted_per_second": 25.172861448422804, "draft_n": 277, "draft_n_accepted": 234 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8447653429602888, "mean_accepted_span_including_target_bonus": 1.8447653429602888, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8447653429602888 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 277.0, "llamacpp:spec_decode_num_accepted_tokens_total": 234.0, "llamacpp:spec_decode_num_drafts_total": 277.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 234.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0179", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 236, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 173.36171905751493, "decode_tps": 26.254920982184306, "first_frame_s": 47.45166747306939, "first_answer_s": 54.433980384026654, "wall_s": 56.40265724307392, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:54:09.984798+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 47253.8, "prompt_per_token_ms": 5.7682861328125, "prompt_per_second": 173.36171905751493, "predicted_n": 236, "predicted_ms": 8950.703, "predicted_per_token_ms": 38.08809787234043, "predicted_per_second": 26.254920982184306, "draft_n": 127, "draft_n_accepted": 108 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8503937007874016, "mean_accepted_span_including_target_bonus": 1.8503937007874016, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8503937007874016 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 127.0, "llamacpp:spec_decode_num_accepted_tokens_total": 108.0, "llamacpp:spec_decode_num_drafts_total": 127.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 108.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0180", "run": "iq3-author32-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "author-cpu-mtp1", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 198, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 169.15165109366444, "decode_tps": 21.604954180498314, "first_frame_s": 97.05843690701295, "first_answer_s": 103.7428842659574, "wall_s": 106.17716595693491, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "48", "mtp_draft_max": 1, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T08:55:56.238012+00:00", "configuration_ids": [ "C034" ], "missing_configuration_match": false, "source": { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 96859.829, "prompt_per_token_ms": 5.911854797363281, "prompt_per_second": 169.15165109366444, "predicted_n": 198, "predicted_ms": 9118.279, "predicted_per_token_ms": 46.285680203045686, "predicted_per_second": 21.604954180498314, "draft_n": 103, "draft_n_accepted": 95 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9223300970873787, "mean_accepted_span_including_target_bonus": 1.9223300970873787, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9223300970873787 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 103.0, "llamacpp:spec_decode_num_accepted_tokens_total": 95.0, "llamacpp:spec_decode_num_drafts_total": 103.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 95.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0181", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 189.8008721943205, "decode_tps": 29.884952873593203, "first_frame_s": 43.16337972600013, "first_answer_s": null, "wall_s": 60.26249382796232, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:02:19.843677+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 43161.024, "prompt_per_token_ms": 5.2686796875, "prompt_per_second": 189.8008721943205, "predicted_n": 512, "predicted_ms": 17098.906, "predicted_per_token_ms": 33.46165557729941, "predicted_per_second": 29.884952873593203 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0182", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.539484736240585, "decode_tps": 29.887586994957534, "first_frame_s": 0.14737643604166806, "first_answer_s": null, "wall_s": 17.245108678005636, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:02:37.130412+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 145.246, "prompt_per_token_ms": 36.3115, "prompt_per_second": 27.539484736240585, "predicted_n": 512, "predicted_ms": 17097.399, "predicted_per_token_ms": 33.458706457925636, "predicted_per_second": 29.887586994957534 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0183", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 193.12725815614343, "decode_tps": 29.66557664540034, "first_frame_s": 42.42007748794276, "first_answer_s": null, "wall_s": 59.64565540594049, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:03:36.820220+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 42417.627, "prompt_per_token_ms": 5.1779329833984375, "prompt_per_second": 193.12725815614343, "predicted_n": 512, "predicted_ms": 17225.352, "predicted_per_token_ms": 33.70910371819961, "predicted_per_second": 29.66557664540034 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0184", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.048044088311865, "decode_tps": 29.835659814157122, "first_frame_s": 0.14972178707830608, "first_answer_s": null, "wall_s": 17.277175244060345, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:03:54.138285+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 147.885, "prompt_per_token_ms": 36.97125, "prompt_per_second": 27.048044088311865, "predicted_n": 512, "predicted_ms": 17127.156, "predicted_per_token_ms": 33.51693933463796, "predicted_per_second": 29.835659814157122 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0185", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 192.69583066628067, "decode_tps": 29.910749484361506, "first_frame_s": 42.514497421099804, "first_answer_s": null, "wall_s": 59.59894419100601, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:04:53.778845+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 42512.596, "prompt_per_token_ms": 5.18952587890625, "prompt_per_second": 192.69583066628067, "predicted_n": 512, "predicted_ms": 17084.159, "predicted_per_token_ms": 33.43279647749511, "predicted_per_second": 29.910749484361506 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0186", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 26.9045024684881, "decode_tps": 29.63420889565598, "first_frame_s": 0.15085649897810072, "first_answer_s": null, "wall_s": 17.39475271489937, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:05:11.216245+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 148.674, "prompt_per_token_ms": 37.1685, "prompt_per_second": 26.9045024684881, "predicted_n": 512, "predicted_ms": 17243.585, "predicted_per_token_ms": 33.74478473581213, "predicted_per_second": 29.63420889565598 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0187", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 262, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 190.63259980816196, "decode_tps": 30.1485700741135, "first_frame_s": 43.15426748304162, "first_answer_s": 49.776629458065145, "wall_s": 51.81165437900927, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:06:03.087716+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 42972.713, "prompt_per_token_ms": 5.245692504882813, "prompt_per_second": 190.63259980816196, "predicted_n": 262, "predicted_ms": 8657.127, "predicted_per_token_ms": 33.16906896551724, "predicted_per_second": 30.1485700741135 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0188", "run": "iq3-cache96-accept-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "cache96-control", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 192, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 186.2479234959372, "decode_tps": 28.501315761003784, "first_frame_s": 88.15405600098893, "first_answer_s": 92.83822556002997, "wall_s": 94.85590794798918, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "6", "cpu_moe": "99", "batch": "2048", "ubatch": "512", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:07:38.016814+00:00", "configuration_ids": [ "C035" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 87968.766, "prompt_per_token_ms": 5.369187377929688, "prompt_per_second": 186.2479234959372, "predicted_n": 192, "predicted_ms": 6701.445, "predicted_per_token_ms": 35.08609947643979, "predicted_per_second": 28.501315761003784 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0189", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 572.7643957569849, "decode_tps": 31.85751176125095, "first_frame_s": 14.304958252003416, "first_answer_s": null, "wall_s": 30.34534983302001, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:13:43.391143+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 14302.565, "prompt_per_token_ms": 1.7459185791015626, "prompt_per_second": 572.7643957569849, "predicted_n": 512, "predicted_ms": 16040.173, "predicted_per_token_ms": 31.389771037181998, "predicted_per_second": 31.85751176125095 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0190", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.673150249996294, "decode_tps": 31.71987391008713, "first_frame_s": 0.13673213392030448, "first_answer_s": null, "wall_s": 16.24669192591682, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:13:59.679260+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 134.802, "prompt_per_token_ms": 33.7005, "prompt_per_second": 29.673150249996294, "predicted_n": 512, "predicted_ms": 16109.774, "predicted_per_token_ms": 31.52597651663405, "predicted_per_second": 31.71987391008713 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0191", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 596.8754241862927, "decode_tps": 31.870134503193874, "first_frame_s": 13.726961942040361, "first_answer_s": null, "wall_s": 29.760915806982666, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:14:29.482444+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 13724.807, "prompt_per_token_ms": 1.6753914794921876, "prompt_per_second": 596.8754241862927, "predicted_n": 512, "predicted_ms": 16033.82, "predicted_per_token_ms": 31.3773385518591, "predicted_per_second": 31.870134503193874 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0192", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.752079761224362, "decode_tps": 31.90526172710362, "first_frame_s": 0.12844797200523317, "first_answer_s": null, "wall_s": 16.144800978014246, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:14:45.668868+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 125.976, "prompt_per_token_ms": 31.494, "prompt_per_second": 31.752079761224362, "predicted_n": 512, "predicted_ms": 16016.167, "predicted_per_token_ms": 31.34279256360078, "predicted_per_second": 31.90526172710362 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0193", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 594.2954743949841, "decode_tps": 31.77359590552137, "first_frame_s": 13.786528495955281, "first_answer_s": null, "wall_s": 29.869202586007304, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:15:15.581126+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 13784.389, "prompt_per_token_ms": 1.6826646728515624, "prompt_per_second": 594.2954743949841, "predicted_n": 512, "predicted_ms": 16082.536, "predicted_per_token_ms": 31.472673189823876, "predicted_per_second": 31.77359590552137 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0194", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.61580474078992, "decode_tps": 31.814721026299317, "first_frame_s": 0.1287495179567486, "first_answer_s": null, "wall_s": 16.190649967989884, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:15:31.820837+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 126.519, "prompt_per_token_ms": 31.62975, "prompt_per_second": 31.61580474078992, "predicted_n": 512, "predicted_ms": 16061.747, "predicted_per_token_ms": 31.431990215264186, "predicted_per_second": 31.814721026299317 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0195", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 180, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 590.4809312933775, "decode_tps": 32.85561754886447, "first_frame_s": 14.061678697937168, "first_answer_s": 17.78534967894666, "wall_s": 19.509935476002283, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:15:51.389929+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 13873.437, "prompt_per_token_ms": 1.6935347900390625, "prompt_per_second": 590.4809312933775, "predicted_n": 180, "predicted_ms": 5448.079, "predicted_per_token_ms": 30.436195530726255, "predicted_per_second": 32.85561754886447 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0196", "run": "iq3-cache96-t14-u4096-accept-r1-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "cache96-t14-u4096-control", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 239, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 584.7973880858455, "decode_tps": 30.969513636800734, "first_frame_s": 28.204994508996606, "first_answer_s": 34.067170909955166, "wall_s": 35.8903516610153, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": "99", "batch": "4096", "ubatch": "4096", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": "96", "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default", "time_utc": "2026-10-05T09:16:27.356215+00:00", "configuration_ids": [ "C036" ], "missing_configuration_match": false, "source": { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 28016.541, "prompt_per_token_ms": 1.7099939575195313, "prompt_per_second": 584.7973880858455, "predicted_n": 239, "predicted_ms": 7684.977, "predicted_per_token_ms": 32.289819327731095, "predicted_per_second": 30.969513636800734 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0197", "run": "opt181/results/opt256-cache4g-ple-overlap-workspace", "engine": "OptLlama", "commit": "167742d9dfa01306786e7d3cfd31fcb8a8b185dc", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "opt256-cache4g-ple-overlap-workspace", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 242, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1241.4314407666566, "decode_tps": 32.79933922260667, "first_frame_s": 6.60158063005656, "first_answer_s": 12.435281731071882, "wall_s": 13.949169452069327, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 0, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T20:31:50.631561+00:00", "configuration_ids": [ "C037" ], "missing_configuration_match": false, "source": { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/requests.json", "bytes": 14071, "sha256": "4c0b93e6ac36788f660eaf6e72ec5fbdc2ad3baa70ac79e3b1d8fc0f06eb8b06" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6598.834, "prompt_per_token_ms": 0.805521728515625, "prompt_per_second": 1241.4314407666566, "predicted_n": 242, "predicted_ms": 7347.709, "predicted_per_token_ms": 30.48841908713693, "predicted_per_second": 32.79933922260667 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0198", "run": "opt181/results/opt256-cache4g-ple-overlap-workspace", "engine": "OptLlama", "commit": "167742d9dfa01306786e7d3cfd31fcb8a8b185dc", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "opt256-cache4g-ple-overlap-workspace", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1287.7534832379831, "decode_tps": 35.37718867797039, "first_frame_s": 6.566646049031988, "first_answer_s": null, "wall_s": 21.01113053993322, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": 0, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T20:32:11.706039+00:00", "configuration_ids": [ "C037" ], "missing_configuration_match": false, "source": { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/requests.json", "bytes": 14071, "sha256": "4c0b93e6ac36788f660eaf6e72ec5fbdc2ad3baa70ac79e3b1d8fc0f06eb8b06" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6361.466, "prompt_per_token_ms": 0.776546142578125, "prompt_per_second": 1287.7534832379831, "predicted_n": 512, "predicted_ms": 14444.336, "predicted_per_token_ms": 28.266802348336594, "predicted_per_second": 35.37718867797039 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0199", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 726.5651905561781, "decode_tps": 31.290741877771502, "first_frame_s": 11.277510773972608, "first_answer_s": null, "wall_s": 27.608293263008818, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:55:53.106072+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11274.969, "prompt_per_token_ms": 1.3763389892578124, "prompt_per_second": 726.5651905561781, "predicted_n": 512, "predicted_ms": 16330.709, "predicted_per_token_ms": 31.958334637964775, "predicted_per_second": 31.290741877771502 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0200", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.36725671098951, "decode_tps": 31.27223212441575, "first_frame_s": 0.1223007730441168, "first_answer_s": null, "wall_s": 16.462799318949692, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:56:09.611950+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 119.878, "prompt_per_token_ms": 29.9695, "prompt_per_second": 33.36725671098951, "predicted_n": 512, "predicted_ms": 16340.375, "predicted_per_token_ms": 31.97725048923679, "predicted_per_second": 31.27223212441575 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0201", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 750.6475618065535, "decode_tps": 31.18458331843268, "first_frame_s": 10.915738604031503, "first_answer_s": null, "wall_s": 27.302153026103042, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:56:36.955217+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10913.244, "prompt_per_token_ms": 1.33218310546875, "prompt_per_second": 750.6475618065535, "predicted_n": 512, "predicted_ms": 16386.302, "predicted_per_token_ms": 32.067127201565555, "predicted_per_second": 31.18458331843268 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0202", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.30208913761498, "decode_tps": 31.162378678639527, "first_frame_s": 0.12621351098641753, "first_answer_s": null, "wall_s": 16.524253226001747, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:56:53.521093+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 123.831, "prompt_per_token_ms": 30.95775, "prompt_per_second": 32.30208913761498, "predicted_n": 512, "predicted_ms": 16397.978, "predicted_per_token_ms": 32.089976516634046, "predicted_per_second": 31.162378678639527 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0203", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 745.5191496130981, "decode_tps": 31.19988023153608, "first_frame_s": 10.990756537066773, "first_answer_s": null, "wall_s": 27.369132912019268, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:57:20.932496+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10988.316, "prompt_per_token_ms": 1.34134716796875, "prompt_per_second": 745.5191496130981, "predicted_n": 512, "predicted_ms": 16378.268, "predicted_per_token_ms": 32.05140508806262, "predicted_per_second": 31.19988023153608 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0204", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.2094905263836, "decode_tps": 30.742888977260492, "first_frame_s": 0.12675935099832714, "first_answer_s": null, "wall_s": 16.74864081502892, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:57:37.724328+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 124.187, "prompt_per_token_ms": 31.04675, "prompt_per_second": 32.2094905263836, "predicted_n": 512, "predicted_ms": 16621.73, "predicted_per_token_ms": 32.52784735812133, "predicted_per_second": 30.742888977260492 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0205", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 229, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 742.6863032490079, "decode_tps": 31.69620136318689, "first_frame_s": 11.216915022931062, "first_answer_s": 16.679536222014576, "wall_s": 18.410332764033228, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:57:56.193872+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11030.229, "prompt_per_token_ms": 1.3464635009765624, "prompt_per_second": 742.6863032490079, "predicted_n": 229, "predicted_ms": 7193.291, "predicted_per_token_ms": 31.54952192982456, "predicted_per_second": 31.69620136318689 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0206", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 214, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 729.6313980911482, "decode_tps": 30.016394870324945, "first_frame_s": 22.59520701598376, "first_answer_s": 27.901350928936154, "wall_s": 29.69162068096921, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:58:25.961762+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22455.174, "prompt_per_token_ms": 1.3705550537109374, "prompt_per_second": 729.6313980911482, "predicted_n": 214, "predicted_ms": 7096.122, "predicted_per_token_ms": 33.31512676056338, "predicted_per_second": 30.016394870324945 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0207", "run": "placement-speed175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "placement-control-mmap", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 266, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 684.3060521586225, "decode_tps": 27.773119812190828, "first_frame_s": 42.13567002606578, "first_answer_s": 49.72098646406084, "wall_s": 51.67775071202777, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T11:59:17.736462+00:00", "configuration_ids": [ "C038" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 41899.381, "prompt_per_token_ms": 1.4613344377790178, "prompt_per_second": 684.3060521586225, "predicted_n": 266, "predicted_ms": 9541.6, "predicted_per_token_ms": 36.006037735849056, "predicted_per_second": 27.773119812190828 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0208", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 725.209318762693, "decode_tps": 29.452984304902365, "first_frame_s": 11.299854899989441, "first_answer_s": null, "wall_s": 28.648926606983878, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:10:10.550128+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 11296.049, "prompt_per_token_ms": 1.3789122314453126, "prompt_per_second": 725.209318762693, "predicted_n": 512, "predicted_ms": 17349.685, "predicted_per_token_ms": 33.9524168297456, "predicted_per_second": 29.452984304902365 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0209", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 32.65679342945316, "decode_tps": 31.109813685091364, "first_frame_s": 0.12495146598666906, "first_answer_s": null, "wall_s": 16.550779615994543, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:10:27.142137+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 122.486, "prompt_per_token_ms": 30.6215, "prompt_per_second": 32.65679342945316, "predicted_n": 512, "predicted_ms": 16425.685, "predicted_per_token_ms": 32.144197651663404, "predicted_per_second": 31.109813685091364 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0210", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 831.7612585262636, "decode_tps": 30.380481343352354, "first_frame_s": 9.851425373926759, "first_answer_s": null, "wall_s": 26.671628917916678, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:10:53.855897+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 9848.98, "prompt_per_token_ms": 1.20226806640625, "prompt_per_second": 831.7612585262636, "predicted_n": 512, "predicted_ms": 16820.01, "predicted_per_token_ms": 32.915870841487276, "predicted_per_second": 30.380481343352354 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0211", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.021554819908694, "decode_tps": 30.96991657455995, "first_frame_s": 0.12306263099890202, "first_answer_s": null, "wall_s": 16.623089801054448, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:11:10.519828+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 121.133, "prompt_per_token_ms": 30.28325, "prompt_per_second": 33.021554819908694, "predicted_n": 512, "predicted_ms": 16499.883, "predicted_per_token_ms": 32.28939921722114, "predicted_per_second": 30.96991657455995 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0212", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 792.4139519411384, "decode_tps": 30.265632154586665, "first_frame_s": 10.340138113009743, "first_answer_s": null, "wall_s": 27.22415046696551, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:11:37.786370+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10338.031, "prompt_per_token_ms": 1.2619666748046876, "prompt_per_second": 792.4139519411384, "predicted_n": 512, "predicted_ms": 16883.837, "predicted_per_token_ms": 33.04077690802348, "predicted_per_second": 30.265632154586665 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0213", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 30.548109453876172, "decode_tps": 31.029178965495678, "first_frame_s": 0.13308130006771535, "first_answer_s": null, "wall_s": 16.601647115079686, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:11:54.431976+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 130.941, "prompt_per_token_ms": 32.73525, "prompt_per_second": 30.548109453876172, "predicted_n": 512, "predicted_ms": 16468.37, "predicted_per_token_ms": 32.227729941291585, "predicted_per_second": 31.029178965495678 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0214", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 229, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 813.9603742786933, "decode_tps": 31.2930279133809, "first_frame_s": 10.254268528078683, "first_answer_s": 15.80326902703382, "wall_s": 17.540385066065937, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:12:12.031653+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10064.372, "prompt_per_token_ms": 1.22856103515625, "prompt_per_second": 813.9603742786933, "predicted_n": 229, "predicted_ms": 7285.968, "predicted_per_token_ms": 31.956, "predicted_per_second": 31.2930279133809 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0215", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 214, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 800.4944069215406, "decode_tps": 30.11917010232599, "first_frame_s": 20.614546439028345, "first_answer_s": 25.911517861066386, "wall_s": 27.68680286000017, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:12:39.794996+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 20467.351, "prompt_per_token_ms": 1.2492279663085937, "prompt_per_second": 800.4944069215406, "predicted_n": 214, "predicted_ms": 7071.908, "predicted_per_token_ms": 33.20144600938967, "predicted_per_second": 30.11917010232599 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0216", "run": "placement-speed175-r2-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "placement-none-lazy", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 266, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 754.5884104726197, "decode_tps": 27.7569745945125, "first_frame_s": 38.2417548490921, "first_answer_s": 45.83526298508514, "wall_s": 47.789490277995355, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": "99", "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 80, "mtp_draft_max": null, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:13:27.685822+00:00", "configuration_ids": [ "C041" ], "missing_configuration_match": false, "source": { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 37996.873, "prompt_per_token_ms": 1.3252257603236608, "prompt_per_second": 754.5884104726197, "predicted_n": 266, "predicted_ms": 9547.15, "predicted_per_token_ms": 36.02698113207547, "predicted_per_second": 27.7569745945125 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0217", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1028.023555694042, "decode_tps": 37.1693455751167, "first_frame_s": 7.972295481013134, "first_answer_s": null, "wall_s": 21.719593668007292, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:37:57.308983+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7968.689, "prompt_per_token_ms": 0.9727403564453125, "prompt_per_second": 1028.023555694042, "predicted_n": 512, "predicted_ms": 13747.888, "predicted_per_token_ms": 26.903890410958905, "predicted_per_second": 37.1693455751167, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0218", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.640343845877442, "decode_tps": 38.16759968148354, "first_frame_s": 0.14693467400502414, "first_answer_s": null, "wall_s": 13.535456717945635, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:38:10.887423+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 144.716, "prompt_per_token_ms": 36.179, "prompt_per_second": 27.640343845877442, "predicted_n": 512, "predicted_ms": 13388.319, "predicted_per_token_ms": 26.200232876712327, "predicted_per_second": 38.16759968148354, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0219", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1201.5310133459118, "decode_tps": 38.540589312008244, "first_frame_s": 6.819993457989767, "first_answer_s": null, "wall_s": 20.078901318949647, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:38:31.008065+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6817.968, "prompt_per_token_ms": 0.832271484375, "prompt_per_second": 1201.5310133459118, "predicted_n": 512, "predicted_ms": 13258.749, "predicted_per_token_ms": 25.94667123287671, "predicted_per_second": 38.540589312008244, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0220", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.69121140142518, "decode_tps": 38.29045339568508, "first_frame_s": 0.13663665892090648, "first_answer_s": null, "wall_s": 13.482377997948788, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:38:44.533344+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 134.72, "prompt_per_token_ms": 33.68, "prompt_per_second": 29.69121140142518, "predicted_n": 512, "predicted_ms": 13345.363, "predicted_per_token_ms": 26.11617025440313, "predicted_per_second": 38.29045339568508, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0221", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1204.929148636365, "decode_tps": 38.253490836904035, "first_frame_s": 6.8010237460257486, "first_answer_s": null, "wall_s": 20.159503187052906, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:39:04.734053+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6798.74, "prompt_per_token_ms": 0.82992431640625, "prompt_per_second": 1204.929148636365, "predicted_n": 512, "predicted_ms": 13358.258, "predicted_per_token_ms": 26.141405088062623, "predicted_per_second": 38.253490836904035, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0222", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.363617010343333, "decode_tps": 38.3002312404568, "first_frame_s": 0.13827415800187737, "first_answer_s": null, "wall_s": 13.480342376045883, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:39:18.256538+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 136.223, "prompt_per_token_ms": 34.05575, "prompt_per_second": 29.363617010343333, "predicted_n": 512, "predicted_ms": 13341.956, "predicted_per_token_ms": 26.109502935420743, "predicted_per_second": 38.3002312404568, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0223", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 196, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1181.9590798610159, "decode_tps": 37.85004742902097, "first_frame_s": 7.141398750944063, "first_answer_s": 11.101553050917573, "wall_s": 12.293583140941337, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:39:30.615482+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6930.866, "prompt_per_token_ms": 0.846052978515625, "prompt_per_second": 1181.9590798610159, "predicted_n": 196, "predicted_ms": 5151.909, "predicted_per_token_ms": 26.42004615384615, "predicted_per_second": 37.85004742902097, "draft_n": 148, "draft_n_accepted": 123 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.831081081081081, "mean_accepted_span_including_target_bonus": 2.6621621621621623, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8783783783783784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7837837837837838 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 148.0, "llamacpp:spec_decode_num_accepted_tokens_total": 123.0, "llamacpp:spec_decode_num_drafts_total": 74.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 58.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0224", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 205, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1147.501096443461, "decode_tps": 38.41423767227797, "first_frame_s": 14.428935057949275, "first_answer_s": 18.59904819203075, "wall_s": 19.739802252035588, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:39:50.428925+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14277.982, "prompt_per_token_ms": 0.8714588623046875, "prompt_per_second": 1147.501096443461, "predicted_n": 205, "predicted_ms": 5310.531, "predicted_per_token_ms": 26.032014705882354, "predicted_per_second": 38.41423767227797, "draft_n": 150, "draft_n_accepted": 129 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.86, "mean_accepted_span_including_target_bonus": 2.7199999999999998, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8933333333333333, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8266666666666667 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 150.0, "llamacpp:spec_decode_num_accepted_tokens_total": 129.0, "llamacpp:spec_decode_num_drafts_total": 75.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 67.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 62.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0225", "run": "prefill-dma175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid42-mtp2-pinned", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 197, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1074.3456803150134, "decode_tps": 36.49395838795014, "first_frame_s": 26.953442707075737, "first_answer_s": 31.05904157005716, "wall_s": 32.32486718008295, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "FAIL_NO_RETRY", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:40:22.856331+00:00", "configuration_ids": [ "C042" ], "missing_configuration_match": false, "source": { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 26687.872, "prompt_per_token_ms": 0.9307991071428571, "prompt_per_second": 1074.3456803150134, "predicted_n": 197, "predicted_ms": 5370.752, "predicted_per_token_ms": 27.40179591836735, "predicted_per_second": 36.49395838795014, "draft_n": 144, "draft_n_accepted": 126 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.875, "mean_accepted_span_including_target_bonus": 2.75, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9166666666666666, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8333333333333334 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 144.0, "llamacpp:spec_decode_num_accepted_tokens_total": 126.0, "llamacpp:spec_decode_num_drafts_total": 72.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 66.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 60.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0226", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1147.4949884641173, "decode_tps": 32.076781394412215, "first_frame_s": 7.141381469089538, "first_answer_s": null, "wall_s": 23.072018312988803, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:57:31.550144+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7139.029, "prompt_per_token_ms": 0.8714635009765626, "prompt_per_second": 1147.4949884641173, "predicted_n": 512, "predicted_ms": 15930.526, "predicted_per_token_ms": 31.175197651663403, "predicted_per_second": 32.076781394412215, "draft_n": 453, "draft_n_accepted": 284 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6269315673289183, "mean_accepted_span_including_target_bonus": 2.251101321585903, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7048458149779736, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5462555066079295 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 453.0, "llamacpp:spec_decode_num_accepted_tokens_total": 284.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 160.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0227", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.481964960494672, "decode_tps": 32.18919739354431, "first_frame_s": 0.14760837308131158, "first_answer_s": null, "wall_s": 16.02289537806064, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:57:47.615043+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 145.55, "prompt_per_token_ms": 36.3875, "prompt_per_second": 27.481964960494672, "predicted_n": 512, "predicted_ms": 15874.891, "predicted_per_token_ms": 31.0663228962818, "predicted_per_second": 32.18919739354431, "draft_n": 459, "draft_n_accepted": 281 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6122004357298475, "mean_accepted_span_including_target_bonus": 2.2217391304347824, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6869565217391305, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5347826086956522 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 459.0, "llamacpp:spec_decode_num_accepted_tokens_total": 281.0, "llamacpp:spec_decode_num_drafts_total": 230.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 158.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 123.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0228", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1197.6198475081408, "decode_tps": 33.07168517516085, "first_frame_s": 6.842340909992345, "first_answer_s": null, "wall_s": 22.293849360896274, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:58:09.950643+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6840.234, "prompt_per_token_ms": 0.834989501953125, "prompt_per_second": 1197.6198475081408, "predicted_n": 512, "predicted_ms": 15451.284, "predicted_per_token_ms": 30.23734637964775, "predicted_per_second": 33.07168517516085, "draft_n": 455, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.621978021978022, "mean_accepted_span_including_target_bonus": 2.2412280701754383, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6973684210526315, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.543859649122807 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 455.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 228.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0229", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.689876776979244, "decode_tps": 33.356010198192145, "first_frame_s": 0.14150984992738813, "first_answer_s": null, "wall_s": 15.461347206961364, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:58:25.454886+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 139.422, "prompt_per_token_ms": 34.8555, "prompt_per_second": 28.689876776979244, "predicted_n": 512, "predicted_ms": 15319.578, "predicted_per_token_ms": 29.97960469667319, "predicted_per_second": 33.356010198192145, "draft_n": 453, "draft_n_accepted": 284 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6269315673289183, "mean_accepted_span_including_target_bonus": 2.251101321585903, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7004405286343612, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5506607929515418 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 453.0, "llamacpp:spec_decode_num_accepted_tokens_total": 284.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 125.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0230", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1197.6697487071908, "decode_tps": 32.878562546671795, "first_frame_s": 6.84213482902851, "first_answer_s": null, "wall_s": 22.38439926004503, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:58:47.881975+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6839.949, "prompt_per_token_ms": 0.8349547119140625, "prompt_per_second": 1197.6697487071908, "predicted_n": 512, "predicted_ms": 15542.042, "predicted_per_token_ms": 30.414954990215264, "predicted_per_second": 32.878562546671795, "draft_n": 455, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.621978021978022, "mean_accepted_span_including_target_bonus": 2.2412280701754383, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6973684210526315, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.543859649122807 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 455.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 228.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0231", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.70758454383648, "decode_tps": 33.09826655115739, "first_frame_s": 0.14134085201658309, "first_answer_s": null, "wall_s": 15.580486144986935, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:59:03.503942+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 139.336, "prompt_per_token_ms": 34.834, "prompt_per_second": 28.70758454383648, "predicted_n": 512, "predicted_ms": 15438.875, "predicted_per_token_ms": 30.2130626223092, "predicted_per_second": 33.09826655115739, "draft_n": 453, "draft_n_accepted": 284 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6269315673289183, "mean_accepted_span_including_target_bonus": 2.251101321585903, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7004405286343612, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5506607929515418 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 453.0, "llamacpp:spec_decode_num_accepted_tokens_total": 284.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 125.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0232", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1198.0053328201252, "decode_tps": 33.01405106440272, "first_frame_s": 6.8400718229822814, "first_answer_s": null, "wall_s": 22.318667115061544, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:59:25.864385+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6838.033, "prompt_per_token_ms": 0.8347208251953125, "prompt_per_second": 1198.0053328201252, "predicted_n": 512, "predicted_ms": 15478.258, "predicted_per_token_ms": 30.290133072407045, "predicted_per_second": 33.01405106440272, "draft_n": 455, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.621978021978022, "mean_accepted_span_including_target_bonus": 2.2412280701754383, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6973684210526315, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.543859649122807 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 455.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 228.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0233", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.156431544343288, "decode_tps": 33.07319207434284, "first_frame_s": 0.13937083596829325, "first_answer_s": null, "wall_s": 15.590123258996755, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:59:41.497518+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 137.191, "prompt_per_token_ms": 34.29775, "prompt_per_second": 29.156431544343288, "predicted_n": 512, "predicted_ms": 15450.58, "predicted_per_token_ms": 30.2359686888454, "predicted_per_second": 33.07319207434284, "draft_n": 453, "draft_n_accepted": 284 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6269315673289183, "mean_accepted_span_including_target_bonus": 2.251101321585903, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7004405286343612, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5506607929515418 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 453.0, "llamacpp:spec_decode_num_accepted_tokens_total": 284.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 125.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0234", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1198.0573686963053, "decode_tps": 32.767885317787886, "first_frame_s": 6.839811987942085, "first_answer_s": null, "wall_s": 22.4345314638922, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:00:03.975559+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6837.736, "prompt_per_token_ms": 0.8346845703125, "prompt_per_second": 1198.0573686963053, "predicted_n": 512, "predicted_ms": 15594.537, "predicted_per_token_ms": 30.51768493150685, "predicted_per_second": 32.767885317787886, "draft_n": 455, "draft_n_accepted": 283 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.621978021978022, "mean_accepted_span_including_target_bonus": 2.2412280701754383, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6973684210526315, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.543859649122807 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 455.0, "llamacpp:spec_decode_num_accepted_tokens_total": 283.0, "llamacpp:spec_decode_num_drafts_total": 228.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 124.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0235", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.363401456424715, "decode_tps": 32.68759657953407, "first_frame_s": 0.13824156299233437, "first_answer_s": null, "wall_s": 15.771277757012285, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:00:19.789094+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 136.224, "prompt_per_token_ms": 34.056, "prompt_per_second": 29.363401456424715, "predicted_n": 512, "predicted_ms": 15632.841, "predicted_per_token_ms": 30.59264383561644, "predicted_per_second": 32.68759657953407, "draft_n": 453, "draft_n_accepted": 284 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6269315673289183, "mean_accepted_span_including_target_bonus": 2.251101321585903, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7004405286343612, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5506607929515418 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 453.0, "llamacpp:spec_decode_num_accepted_tokens_total": 284.0, "llamacpp:spec_decode_num_drafts_total": 227.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 159.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 125.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0236", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 196, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1191.4080168146763, "decode_tps": 37.62185864715269, "first_frame_s": 7.076048861024901, "first_answer_s": 11.091928062029183, "wall_s": 12.259443492046557, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:00:32.106655+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6875.898, "prompt_per_token_ms": 0.839343017578125, "prompt_per_second": 1191.4080168146763, "predicted_n": 196, "predicted_ms": 5183.157, "predicted_per_token_ms": 26.580292307692307, "predicted_per_second": 37.62185864715269, "draft_n": 148, "draft_n_accepted": 123 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.831081081081081, "mean_accepted_span_including_target_bonus": 2.6621621621621623, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8783783783783784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7837837837837838 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 148.0, "llamacpp:spec_decode_num_accepted_tokens_total": 123.0, "llamacpp:spec_decode_num_drafts_total": 74.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 58.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0237", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 204, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1149.7055739037044, "decode_tps": 39.57566311168038, "first_frame_s": 14.395844577928074, "first_answer_s": 18.248189979000017, "wall_s": 19.525733685004525, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:00:51.710716+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14250.605, "prompt_per_token_ms": 0.8697879028320312, "prompt_per_second": 1149.7055739037044, "predicted_n": 204, "predicted_ms": 5129.415, "predicted_per_token_ms": 25.268054187192117, "predicted_per_second": 39.57566311168038, "draft_n": 142, "draft_n_accepted": 133 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9366197183098591, "mean_accepted_span_including_target_bonus": 2.873239436619718, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9577464788732394, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.9154929577464789 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 142.0, "llamacpp:spec_decode_num_accepted_tokens_total": 133.0, "llamacpp:spec_decode_num_drafts_total": 71.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 68.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 65.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0238", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp2-q8", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 244, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1068.1424755729279, "decode_tps": 34.7241523341418, "first_frame_s": 27.10487876890693, "first_answer_s": 32.84157319192309, "wall_s": 34.10351249494124, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:01:25.911153+00:00", "configuration_ids": [ "C044" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 26842.861, "prompt_per_token_ms": 0.9362046944754464, "prompt_per_second": 1068.1424755729279, "predicted_n": 244, "predicted_ms": 6998.011, "predicted_per_token_ms": 28.798399176954735, "predicted_per_second": 34.7241523341418, "draft_n": 186, "draft_n_accepted": 151 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8118279569892473, "mean_accepted_span_including_target_bonus": 2.6236559139784945, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8494623655913979, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7741935483870968 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 186.0, "llamacpp:spec_decode_num_accepted_tokens_total": 151.0, "llamacpp:spec_decode_num_drafts_total": 93.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 79.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 72.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0239", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1008.5451797339051, "decode_tps": 32.10954696633902, "first_frame_s": 8.126252539921552, "first_answer_s": null, "wall_s": 24.040597234969027, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:02:38.487663+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8122.591, "prompt_per_token_ms": 0.9915272216796875, "prompt_per_second": 1008.5451797339051, "predicted_n": 512, "predicted_ms": 15914.27, "predicted_per_token_ms": 31.143385518591, "predicted_per_second": 32.10954696633902, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0240", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 24.300450773361845, "decode_tps": 35.36394586894094, "first_frame_s": 0.1670350970234722, "first_answer_s": null, "wall_s": 14.617021050071344, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:02:53.147579+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 164.606, "prompt_per_token_ms": 41.1515, "prompt_per_second": 24.300450773361845, "predicted_n": 512, "predicted_ms": 14449.745, "predicted_per_token_ms": 28.277387475538163, "predicted_per_second": 35.36394586894094, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0241", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1202.3512189437245, "decode_tps": 35.823770045888075, "first_frame_s": 6.8158046890748665, "first_answer_s": null, "wall_s": 21.08023879805114, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:03:14.270794+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6813.317, "prompt_per_token_ms": 0.8317037353515625, "prompt_per_second": 1202.3512189437245, "predicted_n": 512, "predicted_ms": 14264.272, "predicted_per_token_ms": 27.91442661448141, "predicted_per_second": 35.823770045888075, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0242", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.795001187793797, "decode_tps": 35.89694415847482, "first_frame_s": 0.14090288197621703, "first_answer_s": null, "wall_s": 14.376329972990789, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:03:28.690524+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 16, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 138.913, "prompt_per_token_ms": 34.72825, "prompt_per_second": 28.795001187793797, "predicted_n": 512, "predicted_ms": 14235.195, "predicted_per_token_ms": 27.857524461839528, "predicted_per_second": 35.89694415847482, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0243", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1200.2775641867183, "decode_tps": 36.15407522580555, "first_frame_s": 6.8271358830388635, "first_answer_s": null, "wall_s": 20.961272062035277, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:03:49.694115+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 17, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6825.088, "prompt_per_token_ms": 0.833140625, "prompt_per_second": 1200.2775641867183, "predicted_n": 512, "predicted_ms": 14133.953, "predicted_per_token_ms": 27.659399217221136, "predicted_per_second": 36.15407522580555, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0244", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.282342626439746, "decode_tps": 35.76001294358512, "first_frame_s": 0.14376602205447853, "first_answer_s": null, "wall_s": 14.433672141982242, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:04:04.169322+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 18, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 141.431, "prompt_per_token_ms": 35.35775, "prompt_per_second": 28.282342626439746, "predicted_n": 512, "predicted_ms": 14289.704, "predicted_per_token_ms": 27.964195694716242, "predicted_per_second": 35.76001294358512, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0245", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1191.51493589862, "decode_tps": 35.97140280516403, "first_frame_s": 6.877919806982391, "first_answer_s": null, "wall_s": 21.083809430943802, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:04:25.297101+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 19, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6875.281, "prompt_per_token_ms": 0.8392677001953125, "prompt_per_second": 1191.51493589862, "predicted_n": 512, "predicted_ms": 14205.729, "predicted_per_token_ms": 27.799861056751467, "predicted_per_second": 35.97140280516403, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0246", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.91468721537105, "decode_tps": 35.86116661638384, "first_frame_s": 0.14053563901688904, "first_answer_s": null, "wall_s": 14.39013303606771, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:04:39.733280+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 20, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 138.338, "prompt_per_token_ms": 34.5845, "prompt_per_second": 28.91468721537105, "predicted_n": 512, "predicted_ms": 14249.397, "predicted_per_token_ms": 27.885317025440315, "predicted_per_second": 35.86116661638384, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0247", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1197.900398793209, "decode_tps": 36.09974153715384, "first_frame_s": 6.840904098004103, "first_answer_s": null, "wall_s": 20.99630804604385, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:05:00.773905+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 21, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6838.632, "prompt_per_token_ms": 0.8347939453125, "prompt_per_second": 1197.900398793209, "predicted_n": 512, "predicted_ms": 14155.226, "predicted_per_token_ms": 27.70102935420744, "predicted_per_second": 36.09974153715384, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0248", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.00863006744507, "decode_tps": 35.61087281482695, "first_frame_s": 0.13996759604196995, "first_answer_s": null, "wall_s": 14.489744688034989, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:05:15.309796+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 22, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 137.89, "prompt_per_token_ms": 34.4725, "prompt_per_second": 29.00863006744507, "predicted_n": 512, "predicted_ms": 14349.55, "predicted_per_token_ms": 28.081311154598826, "predicted_per_second": 35.61087281482695, "draft_n": 417, "draft_n_accepted": 302 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7242206235011991, "mean_accepted_span_including_target_bonus": 2.444976076555024, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7894736842105263, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6555023923444976 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 417.0, "llamacpp:spec_decode_num_accepted_tokens_total": 302.0, "llamacpp:spec_decode_num_drafts_total": 209.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 165.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 137.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0249", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 239, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1181.6409461666772, "decode_tps": 37.140638959437894, "first_frame_s": 7.18869108799845, "first_answer_s": 12.39295648294501, "wall_s": 13.59702106390614, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:05:28.967381+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 23, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6932.732, "prompt_per_token_ms": 0.84628076171875, "prompt_per_second": 1181.6409461666772, "predicted_n": 239, "predicted_ms": 6408.075, "predicted_per_token_ms": 26.92468487394958, "predicted_per_second": 37.140638959437894, "draft_n": 182, "draft_n_accepted": 149 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8186813186813187, "mean_accepted_span_including_target_bonus": 2.6373626373626373, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8351648351648352, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8021978021978022 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 182.0, "llamacpp:spec_decode_num_accepted_tokens_total": 149.0, "llamacpp:spec_decode_num_drafts_total": 91.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 76.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 73.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0250", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 200, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1140.9595494881107, "decode_tps": 37.46487432699719, "first_frame_s": 14.561999864992686, "first_answer_s": 18.727962682954967, "wall_s": 19.874325676006265, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:05:48.920171+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 24, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14359.843, "prompt_per_token_ms": 0.8764552612304688, "prompt_per_second": 1140.9595494881107, "predicted_n": 200, "predicted_ms": 5311.642, "predicted_per_token_ms": 26.691668341708542, "predicted_per_second": 37.46487432699719, "draft_n": 148, "draft_n_accepted": 125 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8445945945945946, "mean_accepted_span_including_target_bonus": 2.6891891891891895, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8648648648648649, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8243243243243243 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 148.0, "llamacpp:spec_decode_num_accepted_tokens_total": 125.0, "llamacpp:spec_decode_num_drafts_total": 74.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 64.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 61.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0251", "run": "prefill-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp2-f16", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 196, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1068.9652202665993, "decode_tps": 37.43118420749545, "first_frame_s": 27.181387622025795, "first_answer_s": 31.163158738054335, "wall_s": 32.39157001604326, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:06:21.413395+00:00", "configuration_ids": [ "C043" ], "missing_configuration_match": false, "source": { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, "source_row_zero_based": 25, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 26822.201, "prompt_per_token_ms": 0.935484130859375, "prompt_per_second": 1068.9652202665993, "predicted_n": 196, "predicted_ms": 5209.56, "predicted_per_token_ms": 26.71569230769231, "predicted_per_second": 37.43118420749545, "draft_n": 140, "draft_n_accepted": 126 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9, "mean_accepted_span_including_target_bonus": 2.8, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9571428571428572, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8428571428571429 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 140.0, "llamacpp:spec_decode_num_accepted_tokens_total": 126.0, "llamacpp:spec_decode_num_drafts_total": 70.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 67.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 59.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0252", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1006.2681712046389, "decode_tps": 30.761486450408, "first_frame_s": 8.143187882029451, "first_answer_s": null, "wall_s": 24.754845343064517, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:23:34.260138+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8140.971, "prompt_per_token_ms": 0.9937708740234374, "prompt_per_second": 1006.2681712046389, "predicted_n": 512, "predicted_ms": 16611.681, "predicted_per_token_ms": 32.50818199608611, "predicted_per_second": 30.761486450408 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0253", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.540020343313124, "decode_tps": 31.396668831869427, "first_frame_s": 0.12920646206475794, "first_answer_s": null, "wall_s": 16.40486938902177, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:23:50.707316+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 126.823, "prompt_per_token_ms": 31.70575, "prompt_per_second": 31.540020343313124, "predicted_n": 512, "predicted_ms": 16275.612, "predicted_per_token_ms": 31.850512720156555, "predicted_per_second": 31.396668831869427 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0254", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1004.0657800360541, "decode_tps": 30.882236603983785, "first_frame_s": 8.1611699659843, "first_answer_s": null, "wall_s": 24.70798005303368, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:24:15.457435+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8158.828, "prompt_per_token_ms": 0.99595068359375, "prompt_per_second": 1004.0657800360541, "predicted_n": 512, "predicted_ms": 16546.729, "predicted_per_token_ms": 32.38107436399217, "predicted_per_second": 30.882236603983785 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0255", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.438944666906316, "decode_tps": 31.120715428097718, "first_frame_s": 0.12207045697141439, "first_answer_s": null, "wall_s": 16.542141149984673, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:24:32.041298+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 119.621, "prompt_per_token_ms": 29.90525, "prompt_per_second": 33.438944666906316, "predicted_n": 512, "predicted_ms": 16419.931, "predicted_per_token_ms": 32.1329373776908, "predicted_per_second": 31.120715428097718 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0256", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1170.1416197034023, "decode_tps": 31.289730226804007, "first_frame_s": 7.003191646072082, "first_answer_s": null, "wall_s": 23.334454133058898, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:24:55.419170+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7000.862, "prompt_per_token_ms": 0.854597412109375, "prompt_per_second": 1170.1416197034023, "predicted_n": 512, "predicted_ms": 16331.237, "predicted_per_token_ms": 31.959367906066536, "predicted_per_second": 31.289730226804007 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0257", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 33.17712437274499, "decode_tps": 30.922011479872403, "first_frame_s": 0.12326915503945202, "first_answer_s": null, "wall_s": 16.64879686094355, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:25:12.112105+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 120.565, "prompt_per_token_ms": 30.14125, "prompt_per_second": 33.17712437274499, "predicted_n": 512, "predicted_ms": 16525.445, "predicted_per_token_ms": 32.339422700587086, "predicted_per_second": 30.922011479872403 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0258", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 214, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1135.6912605563002, "decode_tps": 30.594142557500827, "first_frame_s": 7.400857761036605, "first_answer_s": 12.72030231403187, "wall_s": 14.362976311007515, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:25:26.534565+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7213.228, "prompt_per_token_ms": 0.88052099609375, "prompt_per_second": 1135.6912605563002, "predicted_n": 214, "predicted_ms": 6962.117, "predicted_per_token_ms": 32.68599530516432, "predicted_per_second": 30.594142557500827 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0259", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 246, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1117.5530934918697, "decode_tps": 29.32099984370112, "first_frame_s": 14.801285075955093, "first_answer_s": 21.456521996995434, "wall_s": 23.157430469989777, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:25:49.768959+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14660.601, "prompt_per_token_ms": 0.8948120727539063, "prompt_per_second": 1117.5530934918697, "predicted_n": 246, "predicted_ms": 8355.786, "predicted_per_token_ms": 34.105248979591835, "predicted_per_second": 29.32099984370112 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0260", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid38", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 244, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1034.1969977194788, "decode_tps": 27.074665911992852, "first_frame_s": 27.958835977013223, "first_answer_s": 35.11432978801895, "wall_s": 36.934473023051396, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 38, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:26:26.803918+00:00", "configuration_ids": [ "C045" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 27723.925, "prompt_per_token_ms": 0.9669337681361607, "prompt_per_second": 1034.1969977194788, "predicted_n": 244, "predicted_ms": 8975.18, "predicted_per_token_ms": 36.934897119341564, "predicted_per_second": 27.074665911992852 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0261", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 953.6714955599329, "decode_tps": 36.85399313376211, "first_frame_s": 8.59247072506696, "first_answer_s": null, "wall_s": 22.458263944019563, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:26:56.933482+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8589.96, "prompt_per_token_ms": 1.0485791015625, "prompt_per_second": 953.6714955599329, "predicted_n": 512, "predicted_ms": 13865.526, "predicted_per_token_ms": 27.134101761252445, "predicted_per_second": 36.85399313376211, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0262", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.292144686027925, "decode_tps": 38.33914522763173, "first_frame_s": 0.14362333295866847, "first_answer_s": null, "wall_s": 13.472178719006479, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:27:10.450420+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 141.382, "prompt_per_token_ms": 35.3455, "prompt_per_second": 28.292144686027925, "predicted_n": 512, "predicted_ms": 13328.414, "predicted_per_token_ms": 26.083001956947165, "predicted_per_second": 38.33914522763173, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0263", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1049.7357144257548, "decode_tps": 38.36649432277244, "first_frame_s": 7.806121687055565, "first_answer_s": null, "wall_s": 21.12519702303689, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:27:31.621199+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7803.869, "prompt_per_token_ms": 0.9526207275390625, "prompt_per_second": 1049.7357144257548, "predicted_n": 512, "predicted_ms": 13318.913, "predicted_per_token_ms": 26.06440900195695, "predicted_per_second": 38.36649432277244, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0264", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.69495853841414, "decode_tps": 37.42258157665071, "first_frame_s": 0.13690608402248472, "first_answer_s": null, "wall_s": 13.792146740946919, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:27:45.455832+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 134.703, "prompt_per_token_ms": 33.67575, "prompt_per_second": 29.69495853841414, "predicted_n": 512, "predicted_ms": 13654.857, "predicted_per_token_ms": 26.721833659491192, "predicted_per_second": 37.42258157665071, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0265", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1051.5136431071303, "decode_tps": 38.0964735414027, "first_frame_s": 7.7931302019860595, "first_answer_s": null, "wall_s": 21.206614474998787, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:28:06.704108+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7790.674, "prompt_per_token_ms": 0.951010009765625, "prompt_per_second": 1051.5136431071303, "predicted_n": 512, "predicted_ms": 13413.315, "predicted_per_token_ms": 26.249148727984345, "predicted_per_second": 38.0964735414027, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0266", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.98907836472609, "decode_tps": 37.79878795338574, "first_frame_s": 0.14021027996204793, "first_answer_s": null, "wall_s": 13.659333426039666, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:28:20.406899+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 137.983, "prompt_per_token_ms": 34.49575, "prompt_per_second": 28.98907836472609, "predicted_n": 512, "predicted_ms": 13518.952, "predicted_per_token_ms": 26.455874755381604, "predicted_per_second": 37.79878795338574, "draft_n": 395, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7924050632911392, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8484848484848485, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7323232323232324 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 395.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 168.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 145.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0267", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 196, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1056.3884744819384, "decode_tps": 38.457700276855995, "first_frame_s": 7.961593801970594, "first_answer_s": 11.835691387066618, "wall_s": 13.03224016004242, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:28:33.501631+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7754.723, "prompt_per_token_ms": 0.9466214599609375, "prompt_per_second": 1056.3884744819384, "predicted_n": 196, "predicted_ms": 5070.506, "predicted_per_token_ms": 26.002594871794873, "predicted_per_second": 38.457700276855995, "draft_n": 148, "draft_n_accepted": 123 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.831081081081081, "mean_accepted_span_including_target_bonus": 2.6621621621621623, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8783783783783784, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7837837837837838 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 148.0, "llamacpp:spec_decode_num_accepted_tokens_total": 123.0, "llamacpp:spec_decode_num_drafts_total": 74.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 58.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0268", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 205, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1025.5873528253353, "decode_tps": 37.84693955546914, "first_frame_s": 16.121202324982733, "first_answer_s": 20.327172516961582, "wall_s": 21.511785523034632, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:28:55.088232+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 16, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 15975.236, "prompt_per_token_ms": 0.975051025390625, "prompt_per_second": 1025.5873528253353, "predicted_n": 205, "predicted_ms": 5390.132, "predicted_per_token_ms": 26.42221568627451, "predicted_per_second": 37.84693955546914, "draft_n": 150, "draft_n_accepted": 129 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.86, "mean_accepted_span_including_target_bonus": 2.7199999999999998, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8933333333333333, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8266666666666667 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 150.0, "llamacpp:spec_decode_num_accepted_tokens_total": 129.0, "llamacpp:spec_decode_num_drafts_total": 75.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 67.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 62.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0269", "run": "prefill-hybrid175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-hybrid42-mtp2", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 197, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 946.0231936036532, "decode_tps": 37.25485686436894, "first_frame_s": 30.57624561106786, "first_answer_s": 34.59019396896474, "wall_s": 35.8378586450126, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 42, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 2, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:29:31.024035+00:00", "configuration_ids": [ "C046" ], "missing_configuration_match": false, "source": { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, "source_row_zero_based": 17, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 30307.925, "prompt_per_token_ms": 1.0570565359933035, "prompt_per_second": 946.0231936036532, "predicted_n": 197, "predicted_ms": 5261.059, "predicted_per_token_ms": 26.84213775510204, "predicted_per_second": 37.25485686436894, "draft_n": 144, "draft_n_accepted": 126 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.875, "mean_accepted_span_including_target_bonus": 2.75, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9166666666666666, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8333333333333334 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 144.0, "llamacpp:spec_decode_num_accepted_tokens_total": 126.0, "llamacpp:spec_decode_num_drafts_total": 72.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 66.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 60.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0270", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1028.7485748278334, "decode_tps": 27.54335828567551, "first_frame_s": 7.965397355030291, "first_answer_s": null, "wall_s": 26.517986943013966, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:47:30.644738+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7963.073, "prompt_per_token_ms": 0.9720548095703125, "prompt_per_second": 1028.7485748278334, "predicted_n": 512, "predicted_ms": 18552.567, "predicted_per_token_ms": 36.30639334637964, "predicted_per_second": 27.54335828567551, "draft_n": 592, "draft_n_accepted": 313 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5287162162162162, "mean_accepted_span_including_target_bonus": 2.580808080808081, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6818181818181818, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5151515151515151, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.3838383838383838 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 592.0, "llamacpp:spec_decode_num_accepted_tokens_total": 313.0, "llamacpp:spec_decode_num_drafts_total": 198.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 135.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 102.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 76.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0271", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.989447978112253, "decode_tps": 30.8201538474529, "first_frame_s": 0.1452832769136876, "first_answer_s": null, "wall_s": 16.72555310698226, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:47:47.413331+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 142.911, "prompt_per_token_ms": 35.72775, "prompt_per_second": 27.989447978112253, "predicted_n": 512, "predicted_ms": 16580.06, "predicted_per_token_ms": 32.446301369863015, "predicted_per_second": 30.8201538474529, "draft_n": 595, "draft_n_accepted": 312 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5243697478991597, "mean_accepted_span_including_target_bonus": 2.567839195979899, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.678391959798995, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5125628140703518, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.3768844221105528 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 595.0, "llamacpp:spec_decode_num_accepted_tokens_total": 312.0, "llamacpp:spec_decode_num_drafts_total": 199.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 135.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 102.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 75.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0272", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1174.1168575920371, "decode_tps": 30.54779854689323, "first_frame_s": 6.979098733048886, "first_answer_s": null, "wall_s": 23.707200219971128, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:48:11.163352+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6977.159, "prompt_per_token_ms": 0.8517039794921875, "prompt_per_second": 1174.1168575920371, "predicted_n": 512, "predicted_ms": 16727.883, "predicted_per_token_ms": 32.73558317025441, "predicted_per_second": 30.54779854689323, "draft_n": 595, "draft_n_accepted": 312 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5243697478991597, "mean_accepted_span_including_target_bonus": 2.567839195979899, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6733668341708543, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5125628140703518, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.38190954773869346 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 595.0, "llamacpp:spec_decode_num_accepted_tokens_total": 312.0, "llamacpp:spec_decode_num_drafts_total": 199.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 134.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 102.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 76.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0273", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.328308416491307, "decode_tps": 30.551394680135267, "first_frame_s": 0.13840010191779584, "first_answer_s": null, "wall_s": 16.864522005897015, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:48:28.071628+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 136.387, "prompt_per_token_ms": 34.09675, "prompt_per_second": 29.328308416491307, "predicted_n": 512, "predicted_ms": 16725.914, "predicted_per_token_ms": 32.73172994129158, "predicted_per_second": 30.551394680135267, "draft_n": 601, "draft_n_accepted": 310 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5158069883527454, "mean_accepted_span_including_target_bonus": 2.54228855721393, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6666666666666666, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5024875621890548, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.373134328358209 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 601.0, "llamacpp:spec_decode_num_accepted_tokens_total": 310.0, "llamacpp:spec_decode_num_drafts_total": 201.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 134.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 101.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 75.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0274", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1196.4218923591259, "decode_tps": 30.46836676973243, "first_frame_s": 6.849071486969478, "first_answer_s": null, "wall_s": 23.62070839491207, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:48:51.735862+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6847.083, "prompt_per_token_ms": 0.8358255615234375, "prompt_per_second": 1196.4218923591259, "predicted_n": 512, "predicted_ms": 16771.493, "predicted_per_token_ms": 32.82092563600783, "predicted_per_second": 30.46836676973243, "draft_n": 595, "draft_n_accepted": 312 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5243697478991597, "mean_accepted_span_including_target_bonus": 2.567839195979899, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6733668341708543, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5125628140703518, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.38190954773869346 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 595.0, "llamacpp:spec_decode_num_accepted_tokens_total": 312.0, "llamacpp:spec_decode_num_drafts_total": 199.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 134.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 102.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 76.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0275", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.48739076866102, "decode_tps": 30.433492040123134, "first_frame_s": 0.14303751708939672, "first_answer_s": null, "wall_s": 16.93380793510005, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:49:08.713833+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 140.413, "prompt_per_token_ms": 35.10325, "prompt_per_second": 28.48739076866102, "predicted_n": 512, "predicted_ms": 16790.712, "predicted_per_token_ms": 32.858536203522505, "predicted_per_second": 30.433492040123134, "draft_n": 601, "draft_n_accepted": 310 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5158069883527454, "mean_accepted_span_including_target_bonus": 2.54228855721393, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.6666666666666666, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5024875621890548, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.373134328358209 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 601.0, "llamacpp:spec_decode_num_accepted_tokens_total": 310.0, "llamacpp:spec_decode_num_drafts_total": 201.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 134.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 101.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 75.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0276", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 188, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1179.9608878687143, "decode_tps": 41.73148622782847, "first_frame_s": 7.149083313066512, "first_answer_s": 10.521031409036368, "wall_s": 11.63022827799432, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:49:20.404495+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6942.603, "prompt_per_token_ms": 0.8474857177734375, "prompt_per_second": 1179.9608878687143, "predicted_n": 188, "predicted_ms": 4481.029, "predicted_per_token_ms": 23.962721925133692, "predicted_per_second": 41.73148622782847, "draft_n": 159, "draft_n_accepted": 136 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8553459119496856, "mean_accepted_span_including_target_bonus": 3.5660377358490565, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8867924528301887, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8490566037735849, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.8301886792452831 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 159.0, "llamacpp:spec_decode_num_accepted_tokens_total": 136.0, "llamacpp:spec_decode_num_drafts_total": 53.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 47.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 45.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 44.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0277", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 391, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1144.2922174789378, "decode_tps": 34.76420110929892, "first_frame_s": 14.46870718197897, "first_answer_s": 24.648329332005233, "wall_s": 25.687561305006966, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:49:46.166422+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14318.021, "prompt_per_token_ms": 0.8739026489257813, "prompt_per_second": 1144.2922174789378, "predicted_n": 391, "predicted_ms": 11218.437, "predicted_per_token_ms": 28.76522307692308, "predicted_per_second": 34.76420110929892, "draft_n": 381, "draft_n_accepted": 264 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.6929133858267716, "mean_accepted_span_including_target_bonus": 3.078740157480315, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7952755905511811, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.6850393700787402, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.5984251968503937 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 381.0, "llamacpp:spec_decode_num_accepted_tokens_total": 264.0, "llamacpp:spec_decode_num_drafts_total": 127.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 101.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 87.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 76.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0278", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp3-q8", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 227, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1069.9745519061737, "decode_tps": 36.43499971867666, "first_frame_s": 27.061304301023483, "first_answer_s": 32.1775369750103, "wall_s": 33.26470238098409, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:50:19.532219+00:00", "configuration_ids": [ "C048" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 26796.899, "prompt_per_token_ms": 0.9346016671316965, "prompt_per_second": 1069.9745519061737, "predicted_n": 227, "predicted_ms": 6202.827, "predicted_per_token_ms": 27.446137168141593, "predicted_per_second": 36.43499971867666, "draft_n": 204, "draft_n_accepted": 158 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7745098039215687, "mean_accepted_span_including_target_bonus": 3.323529411764706, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8823529411764706, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7647058823529411, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.6764705882352942 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 204.0, "llamacpp:spec_decode_num_accepted_tokens_total": 158.0, "llamacpp:spec_decode_num_drafts_total": 68.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 60.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 52.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 46.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0279", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1041.3872620744098, "decode_tps": 31.25552591275769, "first_frame_s": 7.869024949963205, "first_answer_s": null, "wall_s": 24.218227647943422, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:51:32.224778+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7866.43, "prompt_per_token_ms": 0.960257568359375, "prompt_per_second": 1041.3872620744098, "predicted_n": 512, "predicted_ms": 16349.109, "predicted_per_token_ms": 31.994342465753427, "predicted_per_second": 31.25552591275769, "draft_n": 570, "draft_n_accepted": 321 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5631578947368421, "mean_accepted_span_including_target_bonus": 2.689473684210526, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7421052631578947, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5157894736842106, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.43157894736842106 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 570.0, "llamacpp:spec_decode_num_accepted_tokens_total": 321.0, "llamacpp:spec_decode_num_drafts_total": 190.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 141.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 98.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 82.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0280", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.61291730151577, "decode_tps": 32.257034148142225, "first_frame_s": 0.14186408801469952, "first_answer_s": null, "wall_s": 15.983612403040752, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:51:48.249379+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 139.797, "prompt_per_token_ms": 34.94925, "prompt_per_second": 28.61291730151577, "predicted_n": 512, "predicted_ms": 15841.506, "predicted_per_token_ms": 31.000990215264185, "predicted_per_second": 32.257034148142225, "draft_n": 570, "draft_n_accepted": 321 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5631578947368421, "mean_accepted_span_including_target_bonus": 2.689473684210526, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7368421052631579, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5157894736842106, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.4368421052631579 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 570.0, "llamacpp:spec_decode_num_accepted_tokens_total": 321.0, "llamacpp:spec_decode_num_drafts_total": 190.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 140.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 98.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 83.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0281", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1198.8526605397178, "decode_tps": 32.82787612185733, "first_frame_s": 6.835620573954657, "first_answer_s": null, "wall_s": 22.401918457006104, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:52:10.695787+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6833.2, "prompt_per_token_ms": 0.834130859375, "prompt_per_second": 1198.8526605397178, "predicted_n": 512, "predicted_ms": 15566.039, "predicted_per_token_ms": 30.461915851272018, "predicted_per_second": 32.82787612185733, "draft_n": 567, "draft_n_accepted": 322 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5679012345679012, "mean_accepted_span_including_target_bonus": 2.7037037037037037, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.746031746031746, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5185185185185185, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.43915343915343913 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 567.0, "llamacpp:spec_decode_num_accepted_tokens_total": 322.0, "llamacpp:spec_decode_num_drafts_total": 189.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 141.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 98.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 83.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0282", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.167940656240994, "decode_tps": 32.44534673408696, "first_frame_s": 0.13044431200250983, "first_answer_s": null, "wall_s": 15.880346538964659, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:52:26.619882+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 128.337, "prompt_per_token_ms": 32.08425, "prompt_per_second": 31.167940656240994, "predicted_n": 512, "predicted_ms": 15749.562, "predicted_per_token_ms": 30.821060665362037, "predicted_per_second": 32.44534673408696, "draft_n": 570, "draft_n_accepted": 321 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5631578947368421, "mean_accepted_span_including_target_bonus": 2.689473684210526, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7315789473684211, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5157894736842106, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.4421052631578947 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 570.0, "llamacpp:spec_decode_num_accepted_tokens_total": 321.0, "llamacpp:spec_decode_num_drafts_total": 190.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 139.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 98.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 84.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0283", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1198.3775676358478, "decode_tps": 32.56578144460404, "first_frame_s": 6.8380915119778365, "first_answer_s": null, "wall_s": 22.529671381926164, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:52:49.191452+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 13, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6835.909, "prompt_per_token_ms": 0.8344615478515625, "prompt_per_second": 1198.3775676358478, "predicted_n": 512, "predicted_ms": 15691.317, "predicted_per_token_ms": 30.707078277886495, "predicted_per_second": 32.56578144460404, "draft_n": 567, "draft_n_accepted": 322 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5679012345679012, "mean_accepted_span_including_target_bonus": 2.7037037037037037, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.746031746031746, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5185185185185185, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.43915343915343913 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 567.0, "llamacpp:spec_decode_num_accepted_tokens_total": 322.0, "llamacpp:spec_decode_num_drafts_total": 189.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 141.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 98.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 83.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0284", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 28.185488701142216, "decode_tps": 32.32928445423898, "first_frame_s": 0.14413313497789204, "first_answer_s": null, "wall_s": 15.950524042011239, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:53:05.186717+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 14, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 141.917, "prompt_per_token_ms": 35.47925, "prompt_per_second": 28.185488701142216, "predicted_n": 512, "predicted_ms": 15806.103, "predicted_per_token_ms": 30.931708414872798, "predicted_per_second": 32.32928445423898, "draft_n": 570, "draft_n_accepted": 321 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.5631578947368421, "mean_accepted_span_including_target_bonus": 2.689473684210526, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7315789473684211, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.5157894736842106, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.4421052631578947 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 570.0, "llamacpp:spec_decode_num_accepted_tokens_total": 321.0, "llamacpp:spec_decode_num_drafts_total": 190.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 139.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 98.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 84.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0285", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 301, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1179.1734817241786, "decode_tps": 35.7145960911327, "first_frame_s": 7.199095083982684, "first_answer_s": 14.456959035014734, "wall_s": 15.599217753042467, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:53:20.849041+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 15, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6947.239, "prompt_per_token_ms": 0.8480516357421874, "prompt_per_second": 1179.1734817241786, "predicted_n": 301, "predicted_ms": 8399.927, "predicted_per_token_ms": 27.999756666666666, "predicted_per_second": 35.7145960911327, "draft_n": 291, "draft_n_accepted": 206 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7079037800687286, "mean_accepted_span_including_target_bonus": 3.1237113402061856, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7835051546391752, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.7010309278350515, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.6391752577319587 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 291.0, "llamacpp:spec_decode_num_accepted_tokens_total": 206.0, "llamacpp:spec_decode_num_drafts_total": 97.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 76.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 68.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 62.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0286", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 257, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1144.2560949062097, "decode_tps": 39.406137198008146, "first_frame_s": 14.517423492972739, "first_answer_s": 19.96680568100419, "wall_s": 21.014239372918382, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:53:41.939214+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 16, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14318.473, "prompt_per_token_ms": 0.8739302368164062, "prompt_per_second": 1144.2560949062097, "predicted_n": 257, "predicted_ms": 6496.45, "predicted_per_token_ms": 25.3767578125, "predicted_per_second": 39.406137198008146, "draft_n": 225, "draft_n_accepted": 182 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8088888888888889, "mean_accepted_span_including_target_bonus": 3.4266666666666667, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8666666666666667, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8133333333333334, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.7466666666666667 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 225.0, "llamacpp:spec_decode_num_accepted_tokens_total": 182.0, "llamacpp:spec_decode_num_drafts_total": 75.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 65.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 61.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 56.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0287", "run": "prefill-kv175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp3-f16", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 177, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1071.2423620460684, "decode_tps": 38.33367855424531, "first_frame_s": 27.122557181981392, "first_answer_s": 30.633264856995083, "wall_s": 31.71460281102918, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 3, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T12:54:13.752098+00:00", "configuration_ids": [ "C047" ], "missing_configuration_match": false, "source": { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, "source_row_zero_based": 17, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 26765.185, "prompt_per_token_ms": 0.9334955705915179, "prompt_per_second": 1071.2423620460684, "predicted_n": 177, "predicted_ms": 4591.263, "predicted_per_token_ms": 26.08672159090909, "predicted_per_second": 38.33367855424531, "draft_n": 156, "draft_n_accepted": 124 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7948717948717948, "mean_accepted_span_including_target_bonus": 3.3846153846153846, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8461538461538461, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 0.8076923076923077, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 0.7307692307692307 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 156.0, "llamacpp:spec_decode_num_accepted_tokens_total": 124.0, "llamacpp:spec_decode_num_drafts_total": 52.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 44.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"1\"}": 42.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"2\"}": 38.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0288", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1049.6098236919436, "decode_tps": 32.282862058299685, "first_frame_s": 7.807358104037121, "first_answer_s": null, "wall_s": 23.636264145025052, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:13:50.698846+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7804.805, "prompt_per_token_ms": 0.9527349853515625, "prompt_per_second": 1049.6098236919436, "predicted_n": 512, "predicted_ms": 15828.832, "predicted_per_token_ms": 30.976187866927592, "predicted_per_second": 32.282862058299685, "draft_n": 287, "draft_n_accepted": 223 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7770034843205574, "mean_accepted_span_including_target_bonus": 1.7770034843205575, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7770034843205574 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_total": 223.0, "llamacpp:spec_decode_num_drafts_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 223.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0289", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.939671561802967, "decode_tps": 33.73975765610234, "first_frame_s": 0.13575584604404867, "first_answer_s": null, "wall_s": 15.281201531062834, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:14:06.022001+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 133.602, "prompt_per_token_ms": 33.4005, "prompt_per_second": 29.939671561802967, "predicted_n": 512, "predicted_ms": 15145.337, "predicted_per_token_ms": 29.638624266144813, "predicted_per_second": 33.73975765610234, "draft_n": 289, "draft_n_accepted": 221 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7647058823529411, "mean_accepted_span_including_target_bonus": 1.7647058823529411, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7647058823529411 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 289.0, "llamacpp:spec_decode_num_accepted_tokens_total": 221.0, "llamacpp:spec_decode_num_drafts_total": 289.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 221.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0290", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1196.6975122072788, "decode_tps": 34.18713758121787, "first_frame_s": 6.847654271987267, "first_answer_s": null, "wall_s": 21.794931662036106, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:14:27.858820+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6845.506, "prompt_per_token_ms": 0.835633056640625, "prompt_per_second": 1196.6975122072788, "predicted_n": 512, "predicted_ms": 14947.142, "predicted_per_token_ms": 29.25076712328767, "predicted_per_second": 34.18713758121787, "draft_n": 286, "draft_n_accepted": 224 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7832167832167832, "mean_accepted_span_including_target_bonus": 1.7832167832167833, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7832167832167832 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_total": 224.0, "llamacpp:spec_decode_num_drafts_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 224.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0291", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 26.256194820965572, "decode_tps": 33.993813392059025, "first_frame_s": 0.1545300220604986, "first_answer_s": null, "wall_s": 15.186888523050584, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:14:43.086840+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 152.345, "prompt_per_token_ms": 38.08625, "prompt_per_second": 26.256194820965572, "predicted_n": 512, "predicted_ms": 15032.147, "predicted_per_token_ms": 29.417117416829747, "predicted_per_second": 33.993813392059025, "draft_n": 287, "draft_n_accepted": 223 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7770034843205574, "mean_accepted_span_including_target_bonus": 1.7770034843205575, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7770034843205574 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_total": 223.0, "llamacpp:spec_decode_num_drafts_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 223.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0292", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1197.6678226188458, "decode_tps": 34.128685313618575, "first_frame_s": 6.842167061055079, "first_answer_s": null, "wall_s": 21.8150708390167, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:15:04.943368+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6839.96, "prompt_per_token_ms": 0.8349560546875, "prompt_per_second": 1197.6678226188458, "predicted_n": 512, "predicted_ms": 14972.742, "predicted_per_token_ms": 29.300864970645794, "predicted_per_second": 34.128685313618575, "draft_n": 286, "draft_n_accepted": 224 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7832167832167832, "mean_accepted_span_including_target_bonus": 1.7832167832167833, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7832167832167832 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_total": 224.0, "llamacpp:spec_decode_num_drafts_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 224.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0293", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.698486119670047, "decode_tps": 33.97890820780808, "first_frame_s": 0.1371758959721774, "first_answer_s": null, "wall_s": 15.176148687954992, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:15:20.161934+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 134.687, "prompt_per_token_ms": 33.67175, "prompt_per_second": 29.698486119670047, "predicted_n": 512, "predicted_ms": 15038.741, "predicted_per_token_ms": 29.430021526418788, "predicted_per_second": 33.97890820780808, "draft_n": 287, "draft_n_accepted": 223 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7770034843205574, "mean_accepted_span_including_target_bonus": 1.7770034843205575, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7770034843205574 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_total": 223.0, "llamacpp:spec_decode_num_drafts_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 223.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0294", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1201.4746223685477, "decode_tps": 34.145698022930674, "first_frame_s": 6.820544401067309, "first_answer_s": null, "wall_s": 21.785996661055833, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:15:41.990129+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6818.288, "prompt_per_token_ms": 0.832310546875, "prompt_per_second": 1201.4746223685477, "predicted_n": 512, "predicted_ms": 14965.282, "predicted_per_token_ms": 29.28626614481409, "predicted_per_second": 34.145698022930674, "draft_n": 286, "draft_n_accepted": 224 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7832167832167832, "mean_accepted_span_including_target_bonus": 1.7832167832167833, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7832167832167832 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_total": 224.0, "llamacpp:spec_decode_num_drafts_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 224.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0295", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.520948803294534, "decode_tps": 33.984896739451756, "first_frame_s": 0.1376377330161631, "first_answer_s": null, "wall_s": 15.173922382062301, "repeat": 3, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:15:57.209814+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 7, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 135.497, "prompt_per_token_ms": 33.87425, "prompt_per_second": 29.520948803294534, "predicted_n": 512, "predicted_ms": 15036.091, "predicted_per_token_ms": 29.424835616438358, "predicted_per_second": 33.984896739451756, "draft_n": 287, "draft_n_accepted": 223 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7770034843205574, "mean_accepted_span_including_target_bonus": 1.7770034843205575, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7770034843205574 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_total": 223.0, "llamacpp:spec_decode_num_drafts_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 223.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0296", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1201.6616140257615, "decode_tps": 33.968335130131955, "first_frame_s": 6.8196848599473014, "first_answer_s": null, "wall_s": 21.863222911953926, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:16:19.115781+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 8, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6817.227, "prompt_per_token_ms": 0.8321810302734375, "prompt_per_second": 1201.6616140257615, "predicted_n": 512, "predicted_ms": 15043.422, "predicted_per_token_ms": 29.439181996086106, "predicted_per_second": 33.968335130131955, "draft_n": 286, "draft_n_accepted": 224 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7832167832167832, "mean_accepted_span_including_target_bonus": 1.7832167832167833, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7832167832167832 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_total": 224.0, "llamacpp:spec_decode_num_drafts_total": 286.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 224.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0297", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.13264823055578, "decode_tps": 33.87623255456891, "first_frame_s": 0.13969383703079075, "first_answer_s": null, "wall_s": 15.224194134003483, "repeat": 4, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:16:34.382348+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 9, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 137.303, "prompt_per_token_ms": 34.32575, "prompt_per_second": 29.13264823055578, "predicted_n": 512, "predicted_ms": 15084.322, "predicted_per_token_ms": 29.519221135029355, "predicted_per_second": 33.87623255456891, "draft_n": 287, "draft_n_accepted": 223 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.7770034843205574, "mean_accepted_span_including_target_bonus": 1.7770034843205575, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.7770034843205574 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_total": 223.0, "llamacpp:spec_decode_num_drafts_total": 287.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 223.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 5 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0298", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 202, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1182.2355826644894, "decode_tps": 35.53873271000871, "first_frame_s": 7.1785577040864155, "first_answer_s": 11.484611309017055, "wall_s": 12.834469011053443, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:16:47.276929+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 10, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6929.245, "prompt_per_token_ms": 0.8458551025390625, "prompt_per_second": 1182.2355826644894, "predicted_n": 202, "predicted_ms": 5655.801, "predicted_per_token_ms": 28.138313432835822, "predicted_per_second": 35.53873271000871, "draft_n": 106, "draft_n_accepted": 96 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9056603773584906, "mean_accepted_span_including_target_bonus": 1.9056603773584906, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9056603773584906 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 106.0, "llamacpp:spec_decode_num_accepted_tokens_total": 96.0, "llamacpp:spec_decode_num_drafts_total": 106.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 96.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0299", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 225, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1145.4908289070595, "decode_tps": 34.530280281051816, "first_frame_s": 14.49818233505357, "first_answer_s": 19.59647656197194, "wall_s": 20.985586146009155, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:17:08.338759+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 11, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 14303.039, "prompt_per_token_ms": 0.8729882202148438, "prompt_per_second": 1145.4908289070595, "predicted_n": 225, "predicted_ms": 6487.06, "predicted_per_token_ms": 28.960089285714286, "predicted_per_second": 34.530280281051816, "draft_n": 118, "draft_n_accepted": 107 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.9067796610169492, "mean_accepted_span_including_target_bonus": 1.9067796610169492, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.9067796610169492 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 118.0, "llamacpp:spec_decode_num_accepted_tokens_total": 107.0, "llamacpp:spec_decode_num_drafts_total": 118.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 107.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "7090597d962958a448427144076b41f66f4a118608a386fb017cccf2e0f36c02", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0300", "run": "prefill-mtp1-finalist175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "pinned-hybrid43-mtp1-f16", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 224, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1069.834818504008, "decode_tps": 32.74682256699552, "first_frame_s": 27.151763233006932, "first_answer_s": 32.58423223008867, "wall_s": 33.96219397406094, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": "8192", "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": 0, "mtp_draft_max": 1, "load_mode": "none", "thinking": "native default ON", "time_utc": "2026-10-05T13:17:42.399446+00:00", "configuration_ids": [ "C049" ], "missing_configuration_match": false, "source": { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, "source_row_zero_based": 12, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 26800.399, "prompt_per_token_ms": 0.9347237374441965, "prompt_per_second": 1069.834818504008, "predicted_n": 224, "predicted_ms": 6809.821, "predicted_per_token_ms": 30.537313901345293, "predicted_per_second": 32.74682256699552, "draft_n": 119, "draft_n_accepted": 105 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "VERIFIED_COUNTERS", "ratio": 0.8823529411764706, "mean_accepted_span_including_target_bonus": 1.8823529411764706, "unconditional_prefix_acceptance_by_position": { "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 0.8823529411764706 }, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 119.0, "llamacpp:spec_decode_num_accepted_tokens_total": 105.0, "llamacpp:spec_decode_num_drafts_total": 119.0, "llamacpp:spec_decode_num_accepted_tokens_per_pos_total{position=\"0\"}": 105.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0301", "run": "prefill-observe175-20261005/benchmark", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "benchmark", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 16384, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 739.3958255184706, "decode_tps": 29.339542215869216, "first_frame_s": 22.296371476026252, "first_answer_s": null, "wall_s": 39.71353431197349, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T11:40:35.635075+00:00", "configuration_ids": [], "missing_configuration_match": true, "source": { "path": "prefill-observe175-20261005/benchmark/context-32768.jsonl", "bytes": 12459, "sha256": "2ba68e02684518163fcf67ff0d66d96a5d620a8a4440facf5bf32a0703ac3e17" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 16384, "prompt_ms": 22158.632, "prompt_per_token_ms": 1.35245556640625, "prompt_per_second": 739.3958255184706, "predicted_n": 512, "predicted_ms": 17416.768, "predicted_per_token_ms": 34.08369471624266, "predicted_per_second": 29.339542215869216 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 16384, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0302", "run": "prefill-observe175-20261005/benchmark", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "benchmark", "context_tokens": 32768, "input_tokens": 16384, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 16380, "cache_kind": "reuse-requested", "prefill_tps": 30.57005510252432, "decode_tps": 29.360076168426176, "first_frame_s": 0.13446020300034434, "first_answer_s": null, "wall_s": 17.539334325003438, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "NO_SUMMARY_STATUS", "threads": null, "cpu_moe": null, "batch": null, "ubatch": null, "kv_k": null, "kv_v": null, "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": null, "thinking": null, "time_utc": "2026-10-05T11:40:53.212391+00:00", "configuration_ids": [], "missing_configuration_match": true, "source": { "path": "prefill-observe175-20261005/benchmark/context-32768.jsonl", "bytes": 12459, "sha256": "2ba68e02684518163fcf67ff0d66d96a5d620a8a4440facf5bf32a0703ac3e17" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 16380, "prompt_n": 4, "prompt_ms": 130.847, "prompt_per_token_ms": 32.71175, "prompt_per_second": 30.57005510252432, "predicted_n": 512, "predicted_ms": 17404.587, "predicted_per_token_ms": 34.05985714285714, "predicted_per_second": 29.360076168426176 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 16380, "actual_prefix_reuse": true }, "input_token_sha256": "31325a8755dcbdb128ffca0f8cf2abe6a851786b9d79bc5bff2623de745aeebe", "storage_cache": "not-evicted; process/OS warmness reported separately", "complete_native_template": true, "truncated": false } }, { "id": "M0303", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 993.3967973867068, "decode_tps": 28.224591970992854, "first_frame_s": 8.24889518006239, "first_answer_s": null, "wall_s": 26.353711826028302, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:18:57.299793+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 8246.453, "prompt_per_token_ms": 1.0066470947265624, "prompt_per_second": 993.3967973867068, "predicted_n": 512, "predicted_ms": 18104.779, "predicted_per_token_ms": 35.43009589041095, "predicted_per_second": 28.224591970992854 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0304", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 31.182762169072934, "decode_tps": 28.296789128451145, "first_frame_s": 0.1306188440648839, "first_answer_s": null, "wall_s": 18.189398457063362, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:19:15.530977+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 128.276, "prompt_per_token_ms": 32.069, "prompt_per_second": 31.182762169072934, "predicted_n": 512, "predicted_ms": 18058.586, "predicted_per_token_ms": 35.339698630136986, "predicted_per_second": 28.296789128451145 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0305", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1085.4858536361724, "decode_tps": 28.077091230931, "first_frame_s": 7.548896668013185, "first_answer_s": null, "wall_s": 25.748997153015807, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:19:41.320610+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7546.851, "prompt_per_token_ms": 0.9212464599609375, "prompt_per_second": 1085.4858536361724, "predicted_n": 512, "predicted_ms": 18199.891, "predicted_per_token_ms": 35.61622504892368, "predicted_per_second": 28.077091230931 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0306", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 27.042375402255335, "decode_tps": 28.04723153790984, "first_frame_s": 0.15057023195549846, "first_answer_s": null, "wall_s": 18.369989505037665, "repeat": 1, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:19:59.733042+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 147.916, "prompt_per_token_ms": 36.979, "prompt_per_second": 27.042375402255335, "predicted_n": 512, "predicted_ms": 18219.267, "predicted_per_token_ms": 35.65414285714286, "predicted_per_second": 28.04723153790984 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0307", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1085.9922379129011, "decode_tps": 28.181056726757575, "first_frame_s": 7.545515123056248, "first_answer_s": null, "wall_s": 25.678417009068653, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:20:25.453361+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7543.332, "prompt_per_token_ms": 0.92081689453125, "prompt_per_second": 1085.9922379129011, "predicted_n": 512, "predicted_ms": 18132.748, "predicted_per_token_ms": 35.48482974559687, "predicted_per_second": 28.181056726757575 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0308", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 4, "cached_input_tokens": 8188, "cache_kind": "reuse-requested", "prefill_tps": 29.81470162937344, "decode_tps": 28.23174784353093, "first_frame_s": 0.13623512000776827, "first_answer_s": null, "wall_s": 18.236579342046753, "repeat": 2, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:20:43.730201+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 8188, "prompt_n": 4, "prompt_ms": 134.162, "prompt_per_token_ms": 33.5405, "prompt_per_second": 29.81470162937344, "predicted_n": 512, "predicted_ms": 18100.19, "predicted_per_token_ms": 35.42111545988258, "predicted_per_second": 28.23174784353093 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 4, "prompt_tokens_reused": 8188, "actual_prefix_reuse": true }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Native counters are not visible saved tokens. CPU0 is hybrid, not ree. 3 speed repeats; actual input8192/physical32768. Text-only synthetic fixture, no image/audio/UI acceptance." }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0309", "run": "prefill-path175-20261005", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "uncached-cpu99", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 217, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "natural-quality", "prefill_tps": 1078.9320732446313, "decode_tps": 28.177388663114666, "first_frame_s": 7.779851461062208, "first_answer_s": 13.689712277031504, "wall_s": 15.445685312035494, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_QA_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 99, "batch": 8192, "ubatch": "8192", "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": 0, "mtp_draft_max": null, "load_mode": "mmap", "thinking": "native default ON", "time_utc": "2026-10-05T12:20:59.235051+00:00", "configuration_ids": [ "C050" ], "missing_configuration_match": false, "source": { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, "source_row_zero_based": 6, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7592.693, "prompt_per_token_ms": 0.9268424072265625, "prompt_per_second": 1078.9320732446313, "predicted_n": 217, "predicted_ms": 7665.721, "predicted_per_token_ms": 35.489449074074074, "predicted_per_second": 28.177388663114666 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "draft_acceptance": { "status": "NOT_APPLICABLE_CONTROL", "ratio": null, "metric_deltas": { "llamacpp:spec_decode_num_draft_tokens_total": 0.0, "llamacpp:spec_decode_num_accepted_tokens_total": 0.0, "llamacpp:spec_decode_num_drafts_total": 0.0 }, "caveat": "Accepted draft counters precede visible-output EOS/stop/limit; not visible saved tokens. CPU0 is hybrid: borrowed target tensors and operation offload can still use CUDA. One repeat is exploratory, 8192 actual input is 25% of allocated32768, not a full32k proof. Text-only screen with CPU-resident projector; no image/audio/UI acceptance." }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0310", "run": "upstream178/results/control32-noMTP", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "codacus-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1156.42284759527, "decode_tps": 30.142324861590104, "first_frame_s": 7.0861471929820254, "first_answer_s": null, "wall_s": 24.03910477901809, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": "0", "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:38:25.990391+00:00", "configuration_ids": [ "C051" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/control32-noMTP/requests.json", "bytes": 34778, "sha256": "41ee43a7a40ec41fd35382ef0280d587483484892350d003f9d4072c74cd3db0" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7083.914, "prompt_per_token_ms": 0.864735595703125, "prompt_per_second": 1156.42284759527, "predicted_n": 512, "predicted_ms": 16952.906, "predicted_per_token_ms": 33.175941291585126, "predicted_per_second": 30.142324861590104 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0311", "run": "upstream178/results/control32-noMTP", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "codacus-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 195, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1282.3143770890244, "decode_tps": 29.93277469106754, "first_frame_s": 6.579459304921329, "first_answer_s": 11.382721372996457, "wall_s": 13.06079088198021, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": "0", "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:38:39.107102+00:00", "configuration_ids": [ "C051" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/control32-noMTP/requests.json", "bytes": 34778, "sha256": "41ee43a7a40ec41fd35382ef0280d587483484892350d003f9d4072c74cd3db0" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6388.449, "prompt_per_token_ms": 0.7798399658203125, "prompt_per_second": 1282.3143770890244, "predicted_n": 195, "predicted_ms": 6481.19, "predicted_per_token_ms": 33.40819587628866, "predicted_per_second": 29.93277469106754 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0312", "run": "upstream178/results/control32-noMTP", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "codacus-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 512, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1154.1154265970392, "decode_tps": 27.576280254444338, "first_frame_s": 25.080369552015327, "first_answer_s": null, "wall_s": 43.61131375702098, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": "0", "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:39:22.811408+00:00", "configuration_ids": [ "C051" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/control32-noMTP/requests.json", "bytes": 34778, "sha256": "41ee43a7a40ec41fd35382ef0280d587483484892350d003f9d4072c74cd3db0" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 24843.269, "prompt_per_token_ms": 0.8664644601004464, "prompt_per_second": 1154.1154265970392, "predicted_n": 512, "predicted_ms": 18530.418, "predicted_per_token_ms": 36.263048923679065, "predicted_per_second": 27.576280254444338 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "eab266f2d9bc6ee360c81ab6177085a436334063ad5c0251766de3421d1528d4", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0313", "run": "upstream178/results/control32-noMTP", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "codacus-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 221, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1153.500214571477, "decode_tps": 27.680124953116792, "first_frame_s": 25.37417613901198, "first_answer_s": 31.54302254295908, "wall_s": 33.322737572016194, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": "0", "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:39:56.225980+00:00", "configuration_ids": [ "C051" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/control32-noMTP/requests.json", "bytes": 34778, "sha256": "41ee43a7a40ec41fd35382ef0280d587483484892350d003f9d4072c74cd3db0" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 24856.519, "prompt_per_token_ms": 0.8669265834263393, "prompt_per_second": 1153.500214571477, "predicted_n": 221, "predicted_ms": 7947.941, "predicted_per_token_ms": 36.12700454545455, "predicted_per_second": 27.680124953116792 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4064, "truncated": false } }, { "id": "M0314", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu43-f16-u2048-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 818.4849715288723, "decode_tps": 31.894790881585127, "first_frame_s": 10.011002797051333, "first_answer_s": null, "wall_s": 26.032377335010096, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 2048, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:20:21.805630+00:00", "configuration_ids": [ "C052" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 10008.736, "prompt_per_token_ms": 1.22176953125, "prompt_per_second": 818.4849715288723, "predicted_n": 512, "predicted_ms": 16021.425, "predicted_per_token_ms": 31.35308219178082, "predicted_per_second": 31.894790881585127 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0315", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu43-f16-u2048-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 152, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 822.6959642304007, "decode_tps": 32.075737826938855, "first_frame_s": 10.175304400967434, "first_answer_s": 13.326689700013958, "wall_s": 14.883076456957497, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 2048, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:20:36.743333+00:00", "configuration_ids": [ "C052" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 9957.506, "prompt_per_token_ms": 1.215515869140625, "prompt_per_second": 822.6959642304007, "predicted_n": 152, "predicted_ms": 4707.608, "predicted_per_token_ms": 31.176211920529802, "predicted_per_second": 32.075737826938855 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0316", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu43-f16-u2048-noMTP", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 512, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 738.9676288687983, "decode_tps": 29.77714053790569, "first_frame_s": 177.65769112005364, "first_answer_s": null, "wall_s": 194.8209484750405, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 2048, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:23:51.851520+00:00", "configuration_ids": [ "C052" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 177371.775, "prompt_per_token_ms": 1.3532392501831054, "prompt_per_second": 738.9676288687983, "predicted_n": 512, "predicted_ms": 17160.815, "predicted_per_token_ms": 33.58280821917808, "predicted_per_second": 29.77714053790569 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "f80d502c12228c3f16a2ec6a752fccd3523f5ffd3403b839d7767397ddfeb5db", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0317", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu43-f16-u2048-noMTP", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 280, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 736.7119472024303, "decode_tps": 29.714140384296496, "first_frame_s": 179.47210663906299, "first_answer_s": 187.16906347603071, "wall_s": 188.86395260202698, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 2048, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:27:00.997274+00:00", "configuration_ids": [ "C052" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 177914.856, "prompt_per_token_ms": 1.3573826293945312, "prompt_per_second": 736.7119472024303, "predicted_n": 280, "predicted_ms": 9389.469, "predicted_per_token_ms": 33.654010752688166, "predicted_per_second": 29.714140384296496 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "5cd57289870a0f5f9cb10260b7d1f7e5e502736c47eb279841c6245eab081bca", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0318", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu43-f16-u2048-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 512, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 695.0425769809831, "decode_tps": 28.30228233260211, "first_frame_s": 341.14782722794916, "first_answer_s": null, "wall_s": 359.2072641309351, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 2048, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:33:00.692971+00:00", "configuration_ids": [ "C052" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 339445.392, "prompt_per_token_ms": 1.438760779726104, "prompt_per_second": 695.0425769809831, "predicted_n": 512, "predicted_ms": 18055.081, "predicted_per_token_ms": 35.33283953033268, "predicted_per_second": 28.30228233260211 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "6945518a0f22640c75ce148211a7c5e064402e7a69fb5fe526f95d3b5221fef1", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0319", "run": "upstream178/results/upstream256-cpu43-u2048-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu43-f16-u2048-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 223, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 694.2680459842634, "decode_tps": 28.31785078227425, "first_frame_s": 342.5918638560688, "first_answer_s": 348.6850184709765, "wall_s": 350.4366564799566, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 2048, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:38:51.623226+00:00", "configuration_ids": [ "C052" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 339824.08, "prompt_per_token_ms": 1.4403658727837614, "prompt_per_second": 694.2680459842634, "predicted_n": 223, "predicted_ms": 7839.578, "predicted_per_token_ms": 35.31341441441442, "predicted_per_second": 28.31785078227425 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "273a52049c92150ef0914ca9783c589ef5d3fed19543fae15071c8a478002f26", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0320", "run": "upstream178/results/upstream256-cpu46-u4096-deep179", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu46-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 512, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 905.7578305338382, "decode_tps": 27.393347672937796, "first_frame_s": 260.5154114130419, "first_answer_s": null, "wall_s": 279.17367621301673, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 46, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:47:54.913249+00:00", "configuration_ids": [ "C053" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/requests.json", "bytes": 120120, "sha256": "3b7049b12e032ae91f9608158dee9267dcb16187833d5bfbb045f6b123421a8b" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 260476.909, "prompt_per_token_ms": 1.1040478660953084, "prompt_per_second": 905.7578305338382, "predicted_n": 512, "predicted_ms": 18654.164, "predicted_per_token_ms": 36.50521330724071, "predicted_per_second": 27.393347672937796 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "6945518a0f22640c75ce148211a7c5e064402e7a69fb5fe526f95d3b5221fef1", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0321", "run": "upstream178/results/upstream256-cpu46-u4096-deep179", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu46-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 313, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 905.251338345896, "decode_tps": 27.45755695330948, "first_frame_s": 263.1968820040347, "first_answer_s": 272.7694089090219, "wall_s": 274.56435050594155, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 46, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:52:29.971708+00:00", "configuration_ids": [ "C053" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/requests.json", "bytes": 120120, "sha256": "3b7049b12e032ae91f9608158dee9267dcb16187833d5bfbb045f6b123421a8b" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 260622.647, "prompt_per_token_ms": 1.104665585833026, "prompt_per_second": 905.251338345896, "predicted_n": 313, "predicted_ms": 11362.992, "predicted_per_token_ms": 36.41984615384615, "predicted_per_second": 27.45755695330948 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "273a52049c92150ef0914ca9783c589ef5d3fed19543fae15071c8a478002f26", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0322", "run": "upstream178/results/upstream256-cpu46-u4096-short179", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu46-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1066.825301366427, "decode_tps": 30.44763081979501, "first_frame_s": 7.681155384052545, "first_answer_s": null, "wall_s": 24.464166517020203, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 46, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:41:06.074866+00:00", "configuration_ids": [ "C054" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu46-u4096-short179/requests.json", "bytes": 14476, "sha256": "c10d3cfc6128e73ddea302b574fd15edd45492afe5b368cc83d99703d23ab7e3" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7678.858, "prompt_per_token_ms": 0.937360595703125, "prompt_per_second": 1066.825301366427, "predicted_n": 512, "predicted_ms": 16782.915, "predicted_per_token_ms": 32.84327788649706, "predicted_per_second": 30.44763081979501 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0323", "run": "upstream178/results/upstream256-cpu46-u4096-short179", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu46-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 260, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1082.6135072422244, "decode_tps": 30.361704020745364, "first_frame_s": 7.78587994701229, "first_answer_s": 14.638892211019993, "wall_s": 16.31656840001233, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 46, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T19:41:22.447728+00:00", "configuration_ids": [ "C054" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-cpu46-u4096-short179/requests.json", "bytes": 14476, "sha256": "c10d3cfc6128e73ddea302b574fd15edd45492afe5b368cc83d99703d23ab7e3" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7566.874, "prompt_per_token_ms": 0.923690673828125, "prompt_per_second": 1082.6135072422244, "predicted_n": 260, "predicted_ms": 8530.483, "predicted_per_token_ms": 32.9362277992278, "predicted_per_second": 30.361704020745364 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0324", "run": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu48-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1046.628366950563, "decode_tps": 29.549069062650908, "first_frame_s": 7.829548526904546, "first_answer_s": null, "wall_s": 25.12288550392259, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:04:47.691287+00:00", "configuration_ids": [ "C055" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7827.038, "prompt_per_token_ms": 0.955448974609375, "prompt_per_second": 1046.628366950563, "predicted_n": 512, "predicted_ms": 17293.269, "predicted_per_token_ms": 33.84201369863014, "predicted_per_second": 29.549069062650908 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0325", "run": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu48-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 260, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1063.232619125338, "decode_tps": 29.632463633873634, "first_frame_s": 7.923792988061905, "first_answer_s": 14.957516803056933, "wall_s": 16.66430092405062, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:05:04.411377+00:00", "configuration_ids": [ "C055" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7704.805, "prompt_per_token_ms": 0.9405279541015625, "prompt_per_second": 1063.232619125338, "predicted_n": 260, "predicted_ms": 8740.414, "predicted_per_token_ms": 33.746772200772206, "predicted_per_second": 29.632463633873634 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0326", "run": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu48-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 512, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 962.9052209302617, "decode_tps": 28.059612454470397, "first_frame_s": 136.41042512003332, "first_answer_s": null, "wall_s": 154.6242735889973, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:07:39.325481+00:00", "configuration_ids": [ "C055" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 136121.393, "prompt_per_token_ms": 1.0385238113403321, "prompt_per_second": 962.9052209302617, "predicted_n": 512, "predicted_ms": 18211.228, "predicted_per_token_ms": 35.63841095890411, "predicted_per_second": 28.059612454470397 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "f80d502c12228c3f16a2ec6a752fccd3523f5ffd3403b839d7767397ddfeb5db", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0327", "run": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu48-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 342, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 960.8132782412905, "decode_tps": 27.895257806172953, "first_frame_s": 137.9876763500506, "first_answer_s": 148.44930966105312, "wall_s": 150.2146512370091, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:10:09.826474+00:00", "configuration_ids": [ "C055" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 136417.765, "prompt_per_token_ms": 1.0407849502563478, "prompt_per_second": 960.8132782412905, "predicted_n": 342, "predicted_ms": 12224.3, "predicted_per_token_ms": 35.84838709677419, "predicted_per_second": 27.895257806172953 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "5cd57289870a0f5f9cb10260b7d1f7e5e502736c47eb279841c6245eab081bca", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0328", "run": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu48-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 512, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 898.3678122583424, "decode_tps": 26.778338451786517, "first_frame_s": 264.33161456894595, "first_answer_s": null, "wall_s": 283.4186566369608, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:14:53.729874+00:00", "configuration_ids": [ "C055" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 262619.605, "prompt_per_token_ms": 1.1131298187166476, "prompt_per_second": 898.3678122583424, "predicted_n": 512, "predicted_ms": 19082.588, "predicted_per_token_ms": 37.343616438356165, "predicted_per_second": 26.778338451786517 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "6945518a0f22640c75ce148211a7c5e064402e7a69fb5fe526f95d3b5221fef1", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0329", "run": "upstream178/results/upstream256-f16-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu48-f16-u4096-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 271, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 896.9843837371802, "decode_tps": 26.801361707703357, "first_frame_s": 265.8010503030382, "first_answer_s": 274.0276887030341, "wall_s": 275.8816258700099, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:19:30.097650+00:00", "configuration_ids": [ "C055" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 263024.646, "prompt_per_token_ms": 1.1148466106328598, "prompt_per_second": 896.9843837371802, "predicted_n": 271, "predicted_ms": 10074.115, "predicted_per_token_ms": 37.311537037037034, "predicted_per_second": 26.801361707703357 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "273a52049c92150ef0914ca9783c589ef5d3fed19543fae15071c8a478002f26", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0330", "run": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu48-q8_0-u4096-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1046.5060278210663, "decode_tps": 28.913961612783435, "first_frame_s": 7.830459109973162, "first_answer_s": null, "wall_s": 25.503500169026665, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:47:29.046509+00:00", "configuration_ids": [ "C056" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7827.953, "prompt_per_token_ms": 0.9555606689453126, "prompt_per_second": 1046.5060278210663, "predicted_n": 512, "predicted_ms": 17673.123, "predicted_per_token_ms": 34.58536790606654, "predicted_per_second": 28.913961612783435 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0331", "run": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu48-q8_0-u4096-noMTP", "context_tokens": 262144, "input_tokens": 8192, "output_tokens": 196, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1063.2765036789601, "decode_tps": 29.446869994032102, "first_frame_s": 7.881737474002875, "first_answer_s": 12.776213048957288, "wall_s": 14.503936819033697, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:47:43.606876+00:00", "configuration_ids": [ "C056" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 7704.487, "prompt_per_token_ms": 0.9404891357421875, "prompt_per_second": 1063.2765036789601, "predicted_n": 196, "predicted_ms": 6622.096, "predicted_per_token_ms": 33.959466666666664, "predicted_per_second": 29.446869994032102 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0332", "run": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu48-q8_0-u4096-noMTP", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 512, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 960.7238805815349, "decode_tps": 20.84859748506347, "first_frame_s": 136.66840577404946, "first_answer_s": null, "wall_s": 161.18101246200968, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:50:25.077656+00:00", "configuration_ids": [ "C056" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 136430.459, "prompt_per_token_ms": 1.0408817977905274, "prompt_per_second": 960.7238805815349, "predicted_n": 512, "predicted_ms": 24510.042, "predicted_per_token_ms": 47.96485714285715, "predicted_per_second": 20.84859748506347 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "f80d502c12228c3f16a2ec6a752fccd3523f5ffd3403b839d7767397ddfeb5db", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0333", "run": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu48-q8_0-u4096-noMTP", "context_tokens": 262144, "input_tokens": 131072, "output_tokens": 332, "fresh_input_tokens": 131072, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 957.2504127043268, "decode_tps": 20.87982142516956, "first_frame_s": 137.89029463601764, "first_answer_s": 151.37079939397518, "wall_s": 153.74564786197152, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:52:59.105351+00:00", "configuration_ids": [ "C056" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 131072, "prompt_ms": 136925.509, "prompt_per_token_ms": 1.0446587295532226, "prompt_per_second": 957.2504127043268, "predicted_n": 332, "predicted_ms": 15852.626, "predicted_per_token_ms": 47.89312990936556, "predicted_per_second": 20.87982142516956 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 131072, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "5cd57289870a0f5f9cb10260b7d1f7e5e502736c47eb279841c6245eab081bca", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0334", "run": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx262144-cpu48-q8_0-u4096-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 512, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 894.8165416730511, "decode_tps": 16.76547084676553, "first_frame_s": 264.8561946640257, "first_answer_s": null, "wall_s": 295.3398790390929, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:57:54.932934+00:00", "configuration_ids": [ "C056" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "source_row_zero_based": 4, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 263661.867, "prompt_per_token_ms": 1.1175475121752731, "prompt_per_second": 894.8165416730511, "predicted_n": 512, "predicted_ms": 30479.311, "predicted_per_token_ms": 59.646401174168304, "predicted_per_second": 16.76547084676553 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "6945518a0f22640c75ce148211a7c5e064402e7a69fb5fe526f95d3b5221fef1", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0335", "run": "upstream178/results/upstream256-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx262144-cpu48-q8_0-u4096-noMTP", "context_tokens": 262144, "input_tokens": 235929, "output_tokens": 202, "fresh_input_tokens": 235929, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 895.0264976652107, "decode_tps": 16.766839495800284, "first_frame_s": 265.3648010600591, "first_answer_s": 274.3904977940256, "wall_s": 277.3571261090692, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 48, "batch": 8192, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T18:02:32.799445+00:00", "configuration_ids": [ "C056" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, "source_row_zero_based": 5, "details": { "timings": { "cache_n": 0, "prompt_n": 235929, "prompt_ms": 263600.017, "prompt_per_token_ms": 1.1172853570353793, "prompt_per_second": 895.0264976652107, "predicted_n": 202, "predicted_ms": 11987.948, "predicted_per_token_ms": 59.64153233830846, "predicted_per_second": 16.766839495800284 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 235929, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "273a52049c92150ef0914ca9783c589ef5d3fed19543fae15071c8a478002f26", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0336", "run": "upstream178/results/upstream32-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 512, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1334.0103262039559, "decode_tps": 32.13864169773036, "first_frame_s": 6.14376645500306, "first_answer_s": null, "wall_s": 22.043590831919573, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:42:42.474205+00:00", "configuration_ids": [ "C057" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream32-noMTP/requests.json", "bytes": 32699, "sha256": "2e478a9a8ef727b7a0d4966bcfde2dd6b9b62df6a4c61de2344ed8ffef55b0f6" }, "source_row_zero_based": 0, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 6140.882, "prompt_per_token_ms": 0.749619384765625, "prompt_per_second": 1334.0103262039559, "predicted_n": 512, "predicted_ms": 15899.863, "predicted_per_token_ms": 31.115191780821917, "predicted_per_second": 32.13864169773036 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "baa5510a958bdb9972e4e4fc3f6427a40701053042ade79eb9b7e73f1791de31", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0337", "run": "upstream178/results/upstream32-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 8192, "output_tokens": 214, "fresh_input_tokens": 8192, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1368.895884873717, "decode_tps": 32.15136075728681, "first_frame_s": 6.162865885067731, "first_answer_s": 11.25145644810982, "wall_s": 12.787843017023988, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:42:55.318518+00:00", "configuration_ids": [ "C057" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream32-noMTP/requests.json", "bytes": 32699, "sha256": "2e478a9a8ef727b7a0d4966bcfde2dd6b9b62df6a4c61de2344ed8ffef55b0f6" }, "source_row_zero_based": 1, "details": { "timings": { "cache_n": 0, "prompt_n": 8192, "prompt_ms": 5984.385, "prompt_per_token_ms": 0.7305157470703125, "prompt_per_second": 1368.895884873717, "predicted_n": 214, "predicted_ms": 6624.914, "predicted_per_token_ms": 31.10288262910798, "predicted_per_second": 32.15136075728681 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 8192, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "daf04ee570ab372d53ffab80085f88d1f9e2e2bd37a88357b8690e217cd800ec", "complete_native_template": true, "requested_output_budget": 4096, "truncated": false } }, { "id": "M0338", "run": "upstream178/results/upstream32-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "synthetic-speed", "case": "upstream-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 512, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1253.311350261575, "decode_tps": 31.4175706237476, "first_frame_s": 23.109955381951295, "first_answer_s": null, "wall_s": 39.375287982984446, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": null, "natural_stop": false, "finish": "limit", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:43:34.786952+00:00", "configuration_ids": [ "C057" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream32-noMTP/requests.json", "bytes": 32699, "sha256": "2e478a9a8ef727b7a0d4966bcfde2dd6b9b62df6a4c61de2344ed8ffef55b0f6" }, "source_row_zero_based": 2, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 22876.997, "prompt_per_token_ms": 0.7978863351004464, "prompt_per_second": 1253.311350261575, "predicted_n": 512, "predicted_ms": 16264.784, "predicted_per_token_ms": 31.8293228962818, "predicted_per_second": 31.4175706237476 }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "eab266f2d9bc6ee360c81ab6177085a436334063ad5c0251766de3421d1528d4", "complete_native_template": true, "requested_output_budget": 512, "truncated": false } }, { "id": "M0339", "run": "upstream178/results/upstream32-noMTP", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "context-quality", "case": "upstream-ctx32768-cpu43-f16-u8192-noMTP", "context_tokens": 32768, "input_tokens": 28672, "output_tokens": 197, "fresh_input_tokens": 28672, "cached_input_tokens": 0, "cache_kind": "fresh-prompt", "prefill_tps": 1245.4724457248415, "decode_tps": 31.382213586064758, "first_frame_s": 23.511333732982166, "first_answer_s": 28.173069415963255, "wall_s": 29.75749554298818, "repeat": 0, "recorded_samples": 1, "native_samples_tps": null, "native_stddev_tps": null, "quality_accepted": true, "natural_stop": true, "finish": "eos", "run_status": "PASS_FINITE_PROTOCOL_AND_RETRIEVAL_NOT_PRODUCTION_ACCEPTANCE", "threads": "14", "cpu_moe": 43, "batch": 8192, "ubatch": 8192, "kv_k": "f16", "kv_v": "f16", "expert_cache_slots": null, "mtp_draft_max": false, "load_mode": "none", "thinking": "native ON", "time_utc": "2026-10-05T17:44:04.636422+00:00", "configuration_ids": [ "C057" ], "missing_configuration_match": false, "source": { "path": "upstream178/results/upstream32-noMTP/requests.json", "bytes": 32699, "sha256": "2e478a9a8ef727b7a0d4966bcfde2dd6b9b62df6a4c61de2344ed8ffef55b0f6" }, "source_row_zero_based": 3, "details": { "timings": { "cache_n": 0, "prompt_n": 28672, "prompt_ms": 23020.983, "prompt_per_token_ms": 0.8029081682477679, "prompt_per_second": 1245.4724457248415, "predicted_n": 197, "predicted_ms": 6245.576, "predicted_per_token_ms": 31.86518367346939, "predicted_per_second": 31.382213586064758 }, "quality": { "valid_json": true, "exact_values": true, "reasoning_closed": true, "accepted": true }, "measurement_validation": { "valid": true, "problems": [], "prompt_tokens_processed": 28672, "prompt_tokens_reused": 0, "actual_prefix_reuse": false }, "input_token_sha256": "bcf99a7cf280c2f7e86145fa1754a9e2c325e31028a58d1d9fe21a40d8a9cd61", "complete_native_template": true, "requested_output_budget": 4064, "truncated": false } }, { "id": "M0340", "run": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 209.551255, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 208.811, 207.801, 212.041 ], "native_stddev_tps": 2.214676, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:43:22Z", "configuration_ids": [ "C058" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/results.jsonl", "bytes": 2265, "sha256": "7c7c17e0cd1fa69e131cff29b73a6cb3406e3155080768b3e512fba00b9c67f0" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2443497422, "stddev_ns": 25704142, "samples_ns": [ 2451972999, 2463893480, 2414625787 ] } }, { "id": "M0341", "run": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.668742, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 19.7525, 19.5813, 19.6724 ], "native_stddev_tps": 0.085696, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:43:32Z", "configuration_ids": [ "C058" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/results.jsonl", "bytes": 2265, "sha256": "7c7c17e0cd1fa69e131cff29b73a6cb3406e3155080768b3e512fba00b9c67f0" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13015740217, "stddev_ns": 56725262, "samples_ns": [ 12960357963, 13073719293, 13013143395 ] } }, { "id": "M0342", "run": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 211.051625, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 211.039, 209.209, 212.907 ], "native_stddev_tps": 1.848716, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:45:04Z", "configuration_ids": [ "C059" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/results.jsonl", "bytes": 2264, "sha256": "36dd6008ff9c726d1ef1bf218390853667493641c8c0eb89497710de9c556020" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2426070821, "stddev_ns": 21250281, "samples_ns": [ 2426092968, 2447309907, 2404809590 ] } }, { "id": "M0343", "run": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.811095, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 19.8706, 19.745, 19.8177 ], "native_stddev_tps": 0.063082, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:45:14Z", "configuration_ids": [ "C059" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/results.jsonl", "bytes": 2264, "sha256": "36dd6008ff9c726d1ef1bf218390853667493641c8c0eb89497710de9c556020" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 12922139688, "stddev_ns": 41166685, "samples_ns": [ 12883336449, 12965319755, 12917762861 ] } }, { "id": "M0344", "run": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 216.511527, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 215.951, 214.676, 218.907 ], "native_stddev_tps": 2.170721, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:46:57Z", "configuration_ids": [ "C060" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/results.jsonl", "bytes": 2265, "sha256": "9a8a4e39214934adff34bc4755b8cdd4ac245f4cd37f3345f795ac4cdc62d2cd" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2364928081, "stddev_ns": 23625628, "samples_ns": [ 2370906724, 2384990054, 2338887465 ] } }, { "id": "M0345", "run": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.983607, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 20.0606, 19.9503, 19.9399 ], "native_stddev_tps": 0.066898, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:47:07Z", "configuration_ids": [ "C060" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/results.jsonl", "bytes": 2265, "sha256": "9a8a4e39214934adff34bc4755b8cdd4ac245f4cd37f3345f795ac4cdc62d2cd" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 12810595763, "stddev_ns": 42804987, "samples_ns": [ 12761321610, 12831871670, 12838594010 ] } }, { "id": "M0346", "run": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 213.996816, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 214.11, 212.041, 215.84 ], "native_stddev_tps": 1.901938, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:48:46Z", "configuration_ids": [ "C061" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/results.jsonl", "bytes": 2264, "sha256": "51b9db09d6f4c913e4f86a2833a7cddd68cb355e25c9dfd33c8da79ec4faa68f" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2392685063, "stddev_ns": 21283201, "samples_ns": [ 2391294719, 2414629238, 2372131234 ] } }, { "id": "M0347", "run": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 22.956369, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 22.9389, 22.993, 22.9372 ], "native_stddev_tps": 0.031734, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:48:56Z", "configuration_ids": [ "C061" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/results.jsonl", "bytes": 2264, "sha256": "51b9db09d6f4c913e4f86a2833a7cddd68cb355e25c9dfd33c8da79ec4faa68f" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11151603567, "stddev_ns": 15404007, "samples_ns": [ 11160098751, 11133823236, 11160888716 ] } }, { "id": "M0348", "run": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 214.311759, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 214.065, 212.72, 216.15 ], "native_stddev_tps": 1.728328, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:50:35Z", "configuration_ids": [ "C062" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/results.jsonl", "bytes": 2263, "sha256": "dbca6964ce8a5ce3149b84b5c70adb33b614ea59c6e36968f5893a6a02a5c258" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2389146381, "stddev_ns": 19235467, "samples_ns": [ 2391795572, 2406919879, 2368723693 ] } }, { "id": "M0349", "run": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.838247, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 19.7905, 19.8545, 19.8698 ], "native_stddev_tps": 0.042073, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:50:45Z", "configuration_ids": [ "C062" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/results.jsonl", "bytes": 2263, "sha256": "dbca6964ce8a5ce3149b84b5c70adb33b614ea59c6e36968f5893a6a02a5c258" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 12904404904, "stddev_ns": 27396032, "samples_ns": [ 12935516384, 12893808946, 12883889382 ] } }, { "id": "M0350", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b1024-u1024-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 318.052634, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 320.067, 315.714, 318.377 ], "native_stddev_tps": 2.194462, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 1024, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:37:38Z", "configuration_ids": [ "C063" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/results.jsonl", "bytes": 1137, "sha256": "8889121a50e4975cfc60c415e21f5c8619476775f72a6bf05e21d9af74e724e8" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 12878780922, "stddev_ns": 88994440, "samples_ns": [ 12797320422, 12973761866, 12865260479 ] } }, { "id": "M0351", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b1024-u128-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 86.020942, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 87.3182, 85.2444, 85.5003 ], "native_stddev_tps": 1.13073, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 1024, "ubatch": 128, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:30:51Z", "configuration_ids": [ "C064" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/results.jsonl", "bytes": 1136, "sha256": "fa27ba388a0d88ac40739baaea1796a11f01d6083a0b3b9a15973f35f169776d" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 47621758582, "stddev_ns": 621539107, "samples_ns": [ 46908885752, 48050097793, 47906292203 ] } }, { "id": "M0352", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b1024-u256-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 133.101189, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 134.899, 131.584, 132.82 ], "native_stddev_tps": 1.675259, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 1024, "ubatch": 256, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:34:06Z", "configuration_ids": [ "C065" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/results.jsonl", "bytes": 1136, "sha256": "5a7e4df43f0b0b56f751acb431f15e6498d5e7eb27a69bd78ba1ae30e0a3d00e" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 30776819901, "stddev_ns": 386203727, "samples_ns": [ 30363396450, 31128321284, 30838741971 ] } }, { "id": "M0353", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b1024-u512-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 206.95753, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 209.427, 205.387, 206.059 ], "native_stddev_tps": 2.164818, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 1024, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:36:14Z", "configuration_ids": [ "C066" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/results.jsonl", "bytes": 1137, "sha256": "39888789994ee079760655a6a7003a4e91b57ab865e7bb65b9863bda3b947afc" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 19792936212, "stddev_ns": 205930271, "samples_ns": [ 19558131826, 19942856585, 19877820226 ] } }, { "id": "M0354", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u1024-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 318.363482, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 320.725, 316.137, 318.229 ], "native_stddev_tps": 2.297102, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:45:18Z", "configuration_ids": [ "C067" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/results.jsonl", "bytes": 1137, "sha256": "661868b3ce91d08f020e3bae830898d6045fcd588250416e22bbb1ec1d04881c" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 12866243476, "stddev_ns": 92778075, "samples_ns": [ 12771065449, 12956419058, 12871245922 ] } }, { "id": "M0355", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u128-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 86.536258, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 87.7956, 85.6201, 86.1931 ], "native_stddev_tps": 1.1276, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 128, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:38:34Z", "configuration_ids": [ "C069" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/results.jsonl", "bytes": 1136, "sha256": "18bbe5a2addd0cf5b4069582a1bf7734b5912997241e2c76248ea3854459897a" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 47338090796, "stddev_ns": 613546833, "samples_ns": [ 46653830485, 47839219653, 47521222250 ] } }, { "id": "M0356", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u256-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 133.79176, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 135.668, 132.216, 133.491 ], "native_stddev_tps": 1.745147, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 256, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:41:48Z", "configuration_ids": [ "C070" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/results.jsonl", "bytes": 1137, "sha256": "d987ae554284f51ae337a8240dc330b95614803937a9c4e2dfe042800b352d0e" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 30618202059, "stddev_ns": 398102647, "samples_ns": [ 30191432863, 30979528395, 30683644919 ] } }, { "id": "M0357", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 207.364569, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 209.618, 206.007, 206.468 ], "native_stddev_tps": 1.965389, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:43:55Z", "configuration_ids": [ "C071" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/results.jsonl", "bytes": 1137, "sha256": "0acc280b50cfbfbb574c9418f456eca8ffdb35be4929a62da218abc9811fd49c" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 19753828200, "stddev_ns": 186270075, "samples_ns": [ 19540274170, 19882798196, 19838412235 ] } }, { "id": "M0358", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u1024-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 317.176334, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 319.15, 315.425, 316.954 ], "native_stddev_tps": 1.872691, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 1024, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:53:00Z", "configuration_ids": [ "C072" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/results.jsonl", "bytes": 1136, "sha256": "e65b653539096b7ab7b75b8a250ef9e323fbc7d74b98f9c48370c7efa9fea252" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 12914251961, "stddev_ns": 76171522, "samples_ns": [ 12834084338, 12985671886, 12922999660 ] } }, { "id": "M0359", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u128-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 86.396424, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 87.7152, 85.5121, 85.962 ], "native_stddev_tps": 1.164061, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 128, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:46:15Z", "configuration_ids": [ "C073" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/results.jsonl", "bytes": 1135, "sha256": "497cc81965d54ad11a2b30385c5556b0e5ce3c4ef2e5c49c95ba8d3d9f11ecbe" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 47415071140, "stddev_ns": 634743249, "samples_ns": [ 46696566272, 47899671327, 47648975821 ] } }, { "id": "M0360", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u256-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 133.582034, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 135.477, 132.01, 133.259 ], "native_stddev_tps": 1.75612, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 256, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:49:29Z", "configuration_ids": [ "C074" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/results.jsonl", "bytes": 1136, "sha256": "c5c782ca8b6a7f64c8e6035acbc47dfccc1cb1c699487dc94aada1a0fa1655ea" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 30666327186, "stddev_ns": 401771549, "samples_ns": [ 30233876294, 31028017462, 30737087803 ] } }, { "id": "M0361", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u512-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 206.931589, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 209.105, 205.668, 206.021 ], "native_stddev_tps": 1.890404, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:51:36Z", "configuration_ids": [ "C075" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/results.jsonl", "bytes": 1137, "sha256": "4173fb748d85397af9a100546c4ebf763b52de7bd280ca311421c02b81661e68" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 19795077100, "stddev_ns": 179924799, "samples_ns": [ 19588254240, 19915551410, 19881425651 ] } }, { "id": "M0362", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b512-u128-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 79.315744, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 80.2703, 78.6048, 79.0721 ], "native_stddev_tps": 0.859073, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 512, "ubatch": 128, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:23:23Z", "configuration_ids": [ "C076" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/results.jsonl", "bytes": 1135, "sha256": "24fa719439a54421148231b6365df83782ade33fa185293f0c3923d33062a191" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 51645724199, "stddev_ns": 557035469, "samples_ns": [ 51027578389, 52108778367, 51800815843 ] } }, { "id": "M0363", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b512-u256-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 132.752055, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 134.542, 131.271, 132.443 ], "native_stddev_tps": 1.6575, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 512, "ubatch": 256, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:27:20Z", "configuration_ids": [ "C077" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/results.jsonl", "bytes": 1136, "sha256": "460a1232b5ca8ae8d0ccd0a61650911df8f8ab69ae7c1db75378682896b93786" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 30857709303, "stddev_ns": 384007029, "samples_ns": [ 30443915429, 31202600337, 30926612143 ] } }, { "id": "M0364", "run": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b512-u512-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 207.249783, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 209.209, 205.834, 206.706 ], "native_stddev_tps": 1.752227, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 512, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T22:29:27Z", "configuration_ids": [ "C078" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/results.jsonl", "bytes": 1136, "sha256": "5877447e02a26a5bf68db4e07da2b070635dfa0f4be4b64383c04e40fee43c7a" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 19764529821, "stddev_ns": 166513381, "samples_ns": [ 19578476445, 19899566177, 19815546841 ] } }, { "id": "M0365", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t4-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 164.026014, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 159.792, 164.007, 168.279 ], "native_stddev_tps": 4.243404, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 4, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:37:03Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 3122849528, "stddev_ns": 80802361, "samples_ns": [ 3204166263, 3121810765, 3042571556 ] } }, { "id": "M0366", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t4-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 18.682604, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 18.6875, 18.8017, 18.5586 ], "native_stddev_tps": 0.121599, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 4, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:37:16Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13702973652, "stddev_ns": 89225018, "samples_ns": [ 13699020137, 13615791184, 13794109636 ] } }, { "id": "M0367", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t6-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 167.812163, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 166.203, 165.196, 172.037 ], "native_stddev_tps": 3.693743, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 6, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:37:57Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 2, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 3052004792, "stddev_ns": 66407091, "samples_ns": [ 3080561647, 3099356514, 2976096217 ] } }, { "id": "M0368", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t6-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 18.832125, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 18.8694, 18.8562, 18.7708 ], "native_stddev_tps": 0.053506, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 6, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:38:10Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 3, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13593865869, "stddev_ns": 38682006, "samples_ns": [ 13566967271, 13576433998, 13638196338 ] } }, { "id": "M0369", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t8-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 170.039833, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 170.113, 172.748, 167.258 ], "native_stddev_tps": 2.745373, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 8, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:38:51Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 4, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 3011582937, "stddev_ns": 48661380, "samples_ns": [ 3009755550, 2963861054, 3061132209 ] } }, { "id": "M0370", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t8-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 18.470887, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 18.1504, 18.6737, 18.5886 ], "native_stddev_tps": 0.280805, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 8, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:39:03Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 5, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13861800953, "stddev_ns": 212413764, "samples_ns": [ 14104386220, 13709141166, 13771875473 ] } }, { "id": "M0371", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t12-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 171.976209, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 172.284, 171.427, 172.218 ], "native_stddev_tps": 0.477172, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 12, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:39:44Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 6, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2977171297, "stddev_ns": 8273735, "samples_ns": [ 2971838321, 2986702277, 2972973294 ] } }, { "id": "M0372", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t12-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.130292, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 19.0827, 19.1395, 19.1687 ], "native_stddev_tps": 0.04371, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 12, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:39:56Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 7, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13381964560, "stddev_ns": 30596997, "samples_ns": [ 13415283105, 13375482820, 13355127755 ] } }, { "id": "M0373", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 170.447646, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 170.98, 171.279, 169.084 ], "native_stddev_tps": 1.190315, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:40:37Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 8, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 3003952955, "stddev_ns": 21056999, "samples_ns": [ 2994499072, 2989279760, 3028080033 ] } }, { "id": "M0374", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.267753, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 19.308, 19.29, 19.2053 ], "native_stddev_tps": 0.054863, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:40:49Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 9, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13286519718, "stddev_ns": 37887235, "samples_ns": [ 13258776636, 13271096715, 13329685804 ] } }, { "id": "M0375", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t16-b2048-u512-PP512-TG0", "context_tokens": null, "input_tokens": 512, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 171.981368, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 170.155, 169.091, 176.698 ], "native_stddev_tps": 4.119283, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 16, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:41:29Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 10, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 2978191357, "stddev_ns": 70435045, "samples_ns": [ 3009015063, 3027960131, 2897598879 ] } }, { "id": "M0376", "run": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t16-b2048-u512-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 19.103433, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 19.2272, 19.0642, 19.0189 ], "native_stddev_tps": 0.109544, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "PROCESS_PASS_REQUIRES_RESULT_REVIEW", "threads": 16, "cpu_moe": 99, "batch": 2048, "ubatch": 512, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-04T21:41:41Z", "configuration_ids": [ "C079" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, "source_row_zero_based": 11, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": true, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13401025990, "stddev_ns": 76639464, "samples_ns": [ 13314480315, 13428299914, 13460297743 ] } }, { "id": "M0377", "run": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b8192-u8192-PP8192-TG0", "context_tokens": null, "input_tokens": 8192, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 573.181187, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 810.736, 461.667, 447.14 ], "native_stddev_tps": 205.85677, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_PP8192_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 8192, "ubatch": 8192, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T07:39:43Z", "configuration_ids": [ "C080" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/results.jsonl", "bytes": 1141, "sha256": "79b73c0a2b608355a47aa15a708ffe165066e9808e5aee0e332306c3dc2443bd" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 15389884994, "stddev_ns": 1608921217, "samples_ns": [ 10104397867, 17744373612, 18320883504 ] } }, { "id": "M0378", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u2048-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 314.15487, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 312.197, 312.615, 317.653 ], "native_stddev_tps": 3.036887, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:44:50Z", "configuration_ids": [ "C081" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/results.jsonl", "bytes": 2274, "sha256": "2b18017dda26cf600bd5a52c15ddbc61b72ccfe2e26fbb20e7f0a6cbea27556d" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 13038963256, "stddev_ns": 125361250, "samples_ns": [ 13119937383, 13102388706, 12894563681 ] } }, { "id": "M0379", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u2048-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 22.462031, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 22.4408, 22.4686, 22.4767 ], "native_stddev_tps": 0.018838, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:45:42Z", "configuration_ids": [ "C081" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/results.jsonl", "bytes": 2274, "sha256": "2b18017dda26cf600bd5a52c15ddbc61b72ccfe2e26fbb20e7f0a6cbea27556d" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11397015830, "stddev_ns": 9561758, "samples_ns": [ 11407800471, 11393671889, 11389575130 ] } }, { "id": "M0380", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u2048-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 468.642409, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 470.852, 465.591, 469.484 ], "native_stddev_tps": 2.729401, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:46:21Z", "configuration_ids": [ "C082" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/results.jsonl", "bytes": 2270, "sha256": "75e37c24a1918c44ec3fa4f380427beca7fb1ea8013785f0046ecd977eb807c2" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 8740337563, "stddev_ns": 51029148, "samples_ns": [ 8699127731, 8797415450, 8724469509 ] } }, { "id": "M0381", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u2048-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 22.912849, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 22.8139, 23.0107, 22.9139 ], "native_stddev_tps": 0.098407, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:46:56Z", "configuration_ids": [ "C082" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/results.jsonl", "bytes": 2270, "sha256": "75e37c24a1918c44ec3fa4f380427beca7fb1ea8013785f0046ecd977eb807c2" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11172907740, "stddev_ns": 47989929, "samples_ns": [ 11221218604, 11125245555, 11172259062 ] } }, { "id": "M0382", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u4096-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 642.697501, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 643.077, 641.483, 643.533 ], "native_stddev_tps": 1.07637, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:47:34Z", "configuration_ids": [ "C083" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/results.jsonl", "bytes": 2270, "sha256": "254b2c28d7aa3ee4cf4d89817705c962b8d71704c6f5cb5922dad1cca717b117" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 6373150136, "stddev_ns": 10682140, "samples_ns": [ 6369378662, 6385206111, 6364865636 ] } }, { "id": "M0383", "run": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u4096-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 22.940814, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 22.9803, 22.9109, 22.9312 ], "native_stddev_tps": 0.035654, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:48:01Z", "configuration_ids": [ "C083" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/results.jsonl", "bytes": 2270, "sha256": "254b2c28d7aa3ee4cf4d89817705c962b8d71704c6f5cb5922dad1cca717b117" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11159168520, "stddev_ns": 17333276, "samples_ns": [ 11139983200, 11173699306, 11163823054 ] } }, { "id": "M0384", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u2048-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 476.872743, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 479.035, 474.628, 476.955 ], "native_stddev_tps": 2.204499, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:48:39Z", "configuration_ids": [ "C084" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/results.jsonl", "bytes": 2269, "sha256": "76dfec7805cae6e5518ce01791891a0a570db1e154fff436831d51c016651368" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 8589416014, "stddev_ns": 39718263, "samples_ns": [ 8550524861, 8629912289, 8587810894 ] } }, { "id": "M0385", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u2048-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 23.245437, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 23.2656, 23.2429, 23.2278 ], "native_stddev_tps": 0.018989, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:49:14Z", "configuration_ids": [ "C084" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/results.jsonl", "bytes": 2269, "sha256": "76dfec7805cae6e5518ce01791891a0a570db1e154fff436831d51c016651368" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11012918862, "stddev_ns": 8994748, "samples_ns": [ 11003387547, 11014110861, 11021258178 ] } }, { "id": "M0386", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u2048-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 475.827864, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 477.173, 474.132, 476.178 ], "native_stddev_tps": 1.550243, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:49:52Z", "configuration_ids": [ "C085" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/results.jsonl", "bytes": 2270, "sha256": "2025e9cd8dfff3b553dda790ef88aede231b4b2f30e6248540f9b7572f8c14f9" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 8608215967, "stddev_ns": 28075017, "samples_ns": [ 8583888513, 8638936037, 8601823351 ] } }, { "id": "M0387", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u2048-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 23.239645, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 23.2545, 23.2599, 23.2046 ], "native_stddev_tps": 0.030484, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:50:27Z", "configuration_ids": [ "C085" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/results.jsonl", "bytes": 2270, "sha256": "2025e9cd8dfff3b553dda790ef88aede231b4b2f30e6248540f9b7572f8c14f9" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11015671415, "stddev_ns": 14459951, "samples_ns": [ 11008630958, 11006080039, 11032303248 ] } }, { "id": "M0388", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u4096-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 646.80493, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 646.63, 643.379, 650.406 ], "native_stddev_tps": 3.51676, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:51:06Z", "configuration_ids": [ "C086" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/results.jsonl", "bytes": 2268, "sha256": "830767544026a37f5e44b2b28c2a22a7b4d23df73fa35a8f7bb76cd4eecc9bd7" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 6332791395, "stddev_ns": 34418814, "samples_ns": [ 6334383876, 6366386236, 6297604074 ] } }, { "id": "M0389", "run": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u4096-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 23.236342, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 23.2303, 23.2533, 23.2254 ], "native_stddev_tps": 0.014913, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:51:31Z", "configuration_ids": [ "C086" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/results.jsonl", "bytes": 2268, "sha256": "830767544026a37f5e44b2b28c2a22a7b4d23df73fa35a8f7bb76cd4eecc9bd7" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 11017227461, "stddev_ns": 7069281, "samples_ns": [ 11020100522, 11009174824, 11022407038 ] } }, { "id": "M0390", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u2048-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 483.607531, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 484.518, 482.736, 483.569 ], "native_stddev_tps": 0.891547, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:52:10Z", "configuration_ids": [ "C087" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/results.jsonl", "bytes": 2270, "sha256": "db1d56c38bdd6a7dde6dc9df216375fd06a402f9c926d86ab2c3a858ece4a902" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 8469697048, "stddev_ns": 15612323, "samples_ns": [ 8453763854, 8484967386, 8470359904 ] } }, { "id": "M0391", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b2048-u2048-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 23.723289, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 23.7724, 23.7387, 23.6588 ], "native_stddev_tps": 0.058338, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 2048, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:52:44Z", "configuration_ids": [ "C087" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/results.jsonl", "bytes": 2270, "sha256": "db1d56c38bdd6a7dde6dc9df216375fd06a402f9c926d86ab2c3a858ece4a902" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 10791127470, "stddev_ns": 26560706, "samples_ns": [ 10768798544, 10784083819, 10820500047 ] } }, { "id": "M0392", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u2048-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 485.112907, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 486.213, 483.621, 485.505 ], "native_stddev_tps": 1.339402, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:53:22Z", "configuration_ids": [ "C088" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/results.jsonl", "bytes": 2270, "sha256": "ceac905722f6af1cd4d5b1858a2f7cdc8137aed25ef81e1e5d0d31620555d10e" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 8443438173, "stddev_ns": 23338336, "samples_ns": [ 8424299246, 8469437915, 8436577358 ] } }, { "id": "M0393", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u2048-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 23.631269, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 23.7558, 23.6305, 23.5075 ], "native_stddev_tps": 0.124174, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 2048, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:53:56Z", "configuration_ids": [ "C088" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/results.jsonl", "bytes": 2270, "sha256": "ceac905722f6af1cd4d5b1858a2f7cdc8137aed25ef81e1e5d0d31620555d10e" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 10833303521, "stddev_ns": 56923008, "samples_ns": [ 10776294408, 10833476123, 10890140032 ] } }, { "id": "M0394", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u4096-PP4096-TG0", "context_tokens": null, "input_tokens": 4096, "output_tokens": 0, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": 649.263879, "decode_tps": null, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 650.249, 646.63, 650.912 ], "native_stddev_tps": 2.304649, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:54:34Z", "configuration_ids": [ "C089" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/results.jsonl", "bytes": 2269, "sha256": "c5c590ce00a30246e20ae85cf3b26d124faaebc74bdad7f22ac207a14aec41d7" }, "source_row_zero_based": 0, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 6308736096, "stddev_ns": 22435465, "samples_ns": [ 6299124654, 6334376050, 6292707584 ] } }, { "id": "M0395", "run": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277", "engine": "thecodacus", "commit": "27c54b4bbcefadedcec6397477cc2e866c1db716", "quantization": "UD-Q4_K_XL", "class": "native-random", "case": "t14-b4096-u4096-PP0-TG256", "context_tokens": null, "input_tokens": 0, "output_tokens": 256, "fresh_input_tokens": null, "cached_input_tokens": null, "cache_kind": "random-tokens", "prefill_tps": null, "decode_tps": 23.61403, "first_frame_s": null, "first_answer_s": null, "wall_s": null, "repeat": null, "recorded_samples": 3, "native_samples_tps": [ 23.6609, 23.5988, 23.5824 ], "native_stddev_tps": 0.041433, "quality_accepted": null, "natural_stop": false, "finish": null, "run_status": "VALIDATED_STOCK_SCREEN_REQUIRES_OPERATOR_REVIEW", "threads": 14, "cpu_moe": 99, "batch": 4096, "ubatch": 4096, "kv_k": "q8_0", "kv_v": "q8_0", "expert_cache_slots": null, "mtp_draft_max": null, "load_mode": "mmap", "thinking": null, "time_utc": "2026-10-05T06:55:00Z", "configuration_ids": [ "C089" ], "missing_configuration_match": false, "source": { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/results.jsonl", "bytes": 2269, "sha256": "c5c590ce00a30246e20ae85cf3b26d124faaebc74bdad7f22ac207a14aec41d7" }, "source_row_zero_based": 1, "details": { "n_depth": 0, "model_size": 111323630080, "model_n_params": 176943899520, "sched_async_cpu": false, "flash_attn": 1, "lazy_mode": "on", "cpu_mask": "0xffff", "cpu_strict": true, "fit_target": 0, "fit_min_ctx": 0, "backends": "CUDA", "build_commit": "27c54b4", "build_number": 1, "model_type": "qwen4exp A3B Q4_K - Medium", "n_gpu_layers": 99, "no_host": false, "no_kv_offload": false, "no_op_offload": 0, "poll": 50, "split_mode": "layer", "tensor_buft_overrides": "none", "tensor_split": "0.00", "main_gpu": 0, "embeddings": false, "avg_ns": 10841035138, "stddev_ns": 19005874, "samples_ns": [ 10819525064, 10848021894, 10855558457 ] } }, { "id": "M0396", "run": "strata188", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "quantization": "UD-Q4_K_XL", "class": "warmup", "case": "strata188", "context_tokens": 262144, "input_tokens": 65, "fresh_input_tokens": 65, "cached_input_tokens": 0, "output_tokens": 56, "prefill_tps": 46.55826946493804, "decode_tps": 27.2453050501119, "first_frame_s": 1.446992208948359, "wall_s": 3.4550037289736792, "quality_accepted": null, "natural_stop": true, "finish": "stop", "repeat": -1, "recorded_samples": 1, "run_status": "MAX_CONTEXT_THROUGHPUT_MEASURED_PRODUCTION_RESTORED", "thinking": "high", "cache_kind": "fresh-prompt", "configuration_ids": [ "C094" ], "kv_k": "int8", "kv_v": "int8", "mtp_draft_max": 4, "source": { "path": "jarvis-flash174/report/strata188/receipts/results-r3/benchmark/results.json", "bytes": 10102, "sha256": "4f51786ec9a1902e0dfc20283735e541d6ca2999f4b2bef5ac781645b868c5b4" }, "source_row_zero_based": 0, "details": { "usage": { "prompt_tokens": 65, "completion_tokens": 56, "total_tokens": 121, "prompt_tokens_details": { "cached_tokens": 0 } }, "engine": { "projection": null, "time": 1791296780.1056256, "duration_s": 3.5, "finish": "stop", "prompt_tokens": 65, "reused": 0, "output_tokens": 56, "prompt_total": 65, "prompt_read": 65, "reasoning_recoveries": 0, "reasoning_repeat_coverage": 0.0, "engine_generated": 56, "prompt_ms": 1396.1, "decode_ms": 2055.4, "decode_tok_s": 27.2, "hit_rate": 0.42, "pcie_share": 0.318, "ram_blobs": 19533, "file_blobs": 515, "file_mb": 498.2, "drafts_offered": 50, "drafts_accepted": 33 }, "request_sha256": "5697f18d2e83fa36755333beb99f1e165a8300e9386fceff10d03dedf97e57a7", "sampling_temperature": 0, "benchmark_output_cap": 128 } }, { "id": "M0397", "run": "strata188", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "strata188", "context_tokens": 262144, "input_tokens": 260000, "fresh_input_tokens": 260000, "cached_input_tokens": 0, "output_tokens": 512, "prefill_tps": 1230.0384860118604, "decode_tps": 30.397482708463208, "first_frame_s": 211.67140187695622, "wall_s": 228.4814064099919, "quality_accepted": null, "natural_stop": false, "finish": "length", "repeat": 0, "recorded_samples": 1, "run_status": "MAX_CONTEXT_THROUGHPUT_MEASURED_PRODUCTION_RESTORED", "thinking": "high", "cache_kind": "fresh-prompt", "configuration_ids": [ "C094" ], "kv_k": "int8", "kv_v": "int8", "mtp_draft_max": 4, "source": { "path": "jarvis-flash174/report/strata188/receipts/results-r3/benchmark/results.json", "bytes": 10102, "sha256": "4f51786ec9a1902e0dfc20283735e541d6ca2999f4b2bef5ac781645b868c5b4" }, "source_row_zero_based": 1, "details": { "usage": { "prompt_tokens": 260000, "completion_tokens": 512, "total_tokens": 260512, "prompt_tokens_details": { "cached_tokens": 0 } }, "engine": { "projection": null, "time": 1791296788.2883406, "duration_s": 228.3, "finish": "length", "prompt_tokens": 260000, "reused": 0, "output_tokens": 512, "prompt_total": 260000, "prompt_read": 260000, "reasoning_recoveries": 0, "reasoning_repeat_coverage": 0.0, "engine_generated": 512, "prompt_ms": 211375.5, "decode_ms": 16843.5, "decode_tok_s": 30.4, "hit_rate": 0.555, "pcie_share": 0.269, "ram_blobs": 143746, "file_blobs": 4459, "file_mb": 4919.3, "drafts_offered": 426, "drafts_accepted": 321 }, "request_sha256": "04d27a203e42c58123b8be080850e477cd5a6f83ce1113d5d8e72a5e2a11bd6f", "sampling_temperature": 0, "benchmark_output_cap": 512 } }, { "id": "M0398", "run": "strata188", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "strata188", "context_tokens": 262144, "input_tokens": 260000, "fresh_input_tokens": 260000, "cached_input_tokens": 0, "output_tokens": 512, "prefill_tps": 1224.690706717193, "decode_tps": 30.547467901293494, "first_frame_s": 212.61209778301418, "wall_s": 229.32973375206348, "quality_accepted": null, "natural_stop": false, "finish": "length", "repeat": 1, "recorded_samples": 1, "run_status": "MAX_CONTEXT_THROUGHPUT_MEASURED_PRODUCTION_RESTORED", "thinking": "high", "cache_kind": "fresh-prompt", "configuration_ids": [ "C094" ], "kv_k": "int8", "kv_v": "int8", "mtp_draft_max": 4, "source": { "path": "jarvis-flash174/report/strata188/receipts/results-r3/benchmark/results.json", "bytes": 10102, "sha256": "4f51786ec9a1902e0dfc20283735e541d6ca2999f4b2bef5ac781645b868c5b4" }, "source_row_zero_based": 2, "details": { "usage": { "prompt_tokens": 260000, "completion_tokens": 512, "total_tokens": 260512, "prompt_tokens_details": { "cached_tokens": 0 } }, "engine": { "projection": null, "time": 1791297021.358619, "duration_s": 229.1, "finish": "length", "prompt_tokens": 260000, "reused": 0, "output_tokens": 512, "prompt_total": 260000, "prompt_read": 260000, "reasoning_recoveries": 0, "reasoning_repeat_coverage": 0.0, "engine_generated": 512, "prompt_ms": 212298.5, "decode_ms": 16760.8, "decode_tok_s": 30.5, "hit_rate": 0.555, "pcie_share": 0.265, "ram_blobs": 140724, "file_blobs": 5235, "file_mb": 5007.9, "drafts_offered": 421, "drafts_accepted": 321 }, "request_sha256": "f966b56d27f96dbf7b39a4506af32d472a8cacc8a717ebf4f08d3969061fbb78", "sampling_temperature": 0, "benchmark_output_cap": 512 } }, { "id": "M0399", "run": "strata188", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "strata188", "context_tokens": 262144, "input_tokens": 260000, "fresh_input_tokens": 260000, "cached_input_tokens": 0, "output_tokens": 512, "prefill_tps": 1247.3936668863998, "decode_tps": 29.292963967365807, "first_frame_s": 208.74525400507264, "wall_s": 226.1800853490131, "quality_accepted": null, "natural_stop": false, "finish": "length", "repeat": 2, "recorded_samples": 1, "run_status": "MAX_CONTEXT_THROUGHPUT_MEASURED_PRODUCTION_RESTORED", "thinking": "high", "cache_kind": "fresh-prompt", "configuration_ids": [ "C094" ], "kv_k": "int8", "kv_v": "int8", "mtp_draft_max": 4, "source": { "path": "jarvis-flash174/report/strata188/receipts/results-r3/benchmark/results.json", "bytes": 10102, "sha256": "4f51786ec9a1902e0dfc20283735e541d6ca2999f4b2bef5ac781645b868c5b4" }, "source_row_zero_based": 3, "details": { "usage": { "prompt_tokens": 260000, "completion_tokens": 512, "total_tokens": 260512, "prompt_tokens_details": { "cached_tokens": 0 } }, "engine": { "projection": null, "time": 1791297255.344784, "duration_s": 225.9, "finish": "length", "prompt_tokens": 260000, "reused": 0, "output_tokens": 512, "prompt_total": 260000, "prompt_read": 260000, "reasoning_recoveries": 0, "reasoning_repeat_coverage": 0.0, "engine_generated": 512, "prompt_ms": 208434.6, "decode_ms": 17478.6, "decode_tok_s": 29.3, "hit_rate": 0.509, "pcie_share": 0.285, "ram_blobs": 147710, "file_blobs": 5297, "file_mb": 4485.4, "drafts_offered": 394, "drafts_accepted": 306 }, "request_sha256": "2fb52c6cf19079512e8593938510f21791d7ac5925a47b83ef6d0a8637003003", "sampling_temperature": 0, "benchmark_output_cap": 512 } }, { "id": "M0400", "run": "strata188-ram48", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "quantization": "UD-Q4_K_XL", "class": "warmup", "case": "strata188-ram48", "context_tokens": 262144, "input_tokens": 65, "fresh_input_tokens": 65, "cached_input_tokens": 0, "output_tokens": 55, "prefill_tps": 9.41224170636702, "decode_tps": 6.99852394767649, "first_frame_s": 7.045014240080491, "wall_s": 14.767884521046653, "quality_accepted": null, "natural_stop": true, "finish": "stop", "repeat": -1, "recorded_samples": 1, "run_status": "OWNER_INTERRUPTED", "thinking": "high", "cache_kind": "fresh-prompt", "configuration_ids": [ "C095" ], "kv_k": "int8", "kv_v": "int8", "mtp_draft_max": 4, "source": { "path": "jarvis-flash174/report/strata188-ram48/receipts/results-ram48-r2/benchmark/results.json", "bytes": 4324, "sha256": "7e17aab4c46498826c125c80df08681b3b2d1b265c4339f50b201d3584581af1" }, "source_row_zero_based": 0, "details": { "usage": { "prompt_tokens": 65, "completion_tokens": 55, "total_tokens": 120, "prompt_tokens_details": { "cached_tokens": 0 } }, "engine": { "projection": null, "time": 1791298442.7663074, "duration_s": 14.8, "finish": "stop", "prompt_tokens": 65, "reused": 0, "output_tokens": 55, "prompt_total": 65, "prompt_read": 65, "reasoning_recoveries": 0, "reasoning_repeat_coverage": 0.0, "engine_generated": 55, "prompt_ms": 6905.9, "decode_ms": 7858.8, "decode_tok_s": 7.0, "hit_rate": 0.35, "pcie_share": 0.189, "ram_blobs": 11913, "file_blobs": 8071, "file_mb": 23551.1, "drafts_offered": 47, "drafts_accepted": 32 }, "request_sha256": "5697f18d2e83fa36755333beb99f1e165a8300e9386fceff10d03dedf97e57a7", "sampling_temperature": 0, "benchmark_output_cap": 128 } }, { "id": "M0401", "run": "strata188-ram48", "engine": "Strata", "commit": "82f46a8c8f475f001ad76d92f58f4a4f8ffb0253", "quantization": "UD-Q4_K_XL", "class": "synthetic-speed", "case": "strata188-ram48", "context_tokens": 262144, "input_tokens": 260000, "fresh_input_tokens": 260000, "cached_input_tokens": 0, "output_tokens": 512, "prefill_tps": 773.1300437442926, "decode_tps": 9.300399263233995, "first_frame_s": 336.625337178004, "wall_s": 391.5990299190162, "quality_accepted": null, "natural_stop": false, "finish": "length", "repeat": 0, "recorded_samples": 1, "run_status": "OWNER_INTERRUPTED", "thinking": "high", "cache_kind": "fresh-prompt", "configuration_ids": [ "C095" ], "kv_k": "int8", "kv_v": "int8", "mtp_draft_max": 4, "source": { "path": "jarvis-flash174/report/strata188-ram48/receipts/results-ram48-r2/benchmark/results.json", "bytes": 4324, "sha256": "7e17aab4c46498826c125c80df08681b3b2d1b265c4339f50b201d3584581af1" }, "source_row_zero_based": 1, "details": { "usage": { "prompt_tokens": 260000, "completion_tokens": 512, "total_tokens": 260512, "prompt_tokens_details": { "cached_tokens": 0 } }, "engine": { "projection": null, "time": 1791298462.1226792, "duration_s": 391.4, "finish": "length", "prompt_tokens": 260000, "reused": 0, "output_tokens": 512, "prompt_total": 260000, "prompt_read": 260000, "reasoning_recoveries": 0, "reasoning_repeat_coverage": 0.0, "engine_generated": 512, "prompt_ms": 336295.3, "decode_ms": 55051.4, "decode_tok_s": 9.3, "hit_rate": 0.488, "pcie_share": 0.167, "ram_blobs": 88740, "file_blobs": 56564, "file_mb": 161129.5, "drafts_offered": 387, "drafts_accepted": 305 }, "request_sha256": "04d27a203e42c58123b8be080850e477cd5a6f83ce1113d5d8e72a5e2a11bd6f", "sampling_temperature": 0, "benchmark_output_cap": 512 } }, { "id": "M0402", "run": "functional195", "case": "public Responses long-input keepalive", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "functional-transport", "context_tokens": 262144, "input_tokens": 230100, "fresh_input_tokens": 230100, "cached_input_tokens": 0, "output_tokens": 80, "prefill_tps": 1067.78, "decode_tps": 26.56, "quality_accepted": null, "natural_stop": true, "finish": "completed", "run_status": "PASS_TRANSPORT_NOT_BENCHMARK", "recorded_samples": 1, "thinking": "xhigh", "threads": 14, "cpu_moe": 48, "kv_k": "f16", "kv_v": "f16", "batch": 8192, "ubatch": 2048, "configuration_ids": [ "C096" ], "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "source_row_zero_based": 0, "details": { "phase": "public Responses long-input keepalive", "prompt_eval_ms": 215494.3, "eval_ms": 2974.66, "outcome": "natural completed, LONG_STREAM_OK" } }, { "id": "M0403", "run": "functional195", "case": "stock Codex long-input turn", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "functional-transport", "context_tokens": 262144, "input_tokens": 232796, "fresh_input_tokens": 226582, "cached_input_tokens": 6214, "output_tokens": 34, "prefill_tps": 917.53, "decode_tps": 26.93, "quality_accepted": null, "natural_stop": true, "finish": "completed", "run_status": "PASS_TRANSPORT_NOT_BENCHMARK", "recorded_samples": 1, "thinking": "xhigh", "threads": 14, "cpu_moe": 48, "kv_k": "f16", "kv_v": "f16", "batch": 8192, "ubatch": 2048, "configuration_ids": [ "C096" ], "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "source_row_zero_based": 1, "details": { "phase": "stock Codex long-input turn", "prompt_eval_ms": 246947.79, "eval_ms": 1225.4, "outcome": "natural completed, LONG_STREAM_OK" } }, { "id": "M0404", "run": "functional195", "case": "stock native compaction", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "functional-transport", "context_tokens": 262144, "input_tokens": 230530, "fresh_input_tokens": 230530, "cached_input_tokens": 0, "output_tokens": 216, "prefill_tps": 919.76, "decode_tps": 26.77, "quality_accepted": null, "natural_stop": true, "finish": "completed", "run_status": "PASS_TRANSPORT_NOT_BENCHMARK", "recorded_samples": 1, "thinking": "xhigh", "threads": 14, "cpu_moe": 48, "kv_k": "f16", "kv_v": "f16", "batch": 8192, "ubatch": 2048, "configuration_ids": [ "C096" ], "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "source_row_zero_based": 2, "details": { "phase": "stock native compaction", "prompt_eval_ms": 250641.58, "eval_ms": 8030.78, "outcome": "native contextCompaction completed, rollout compacted" } }, { "id": "M0405", "run": "functional195", "case": "after-compaction file tool call", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "functional-transport", "context_tokens": 262144, "input_tokens": 48886, "fresh_input_tokens": 48886, "cached_input_tokens": 0, "output_tokens": 353, "prefill_tps": 862.41, "decode_tps": 28.46, "quality_accepted": null, "natural_stop": true, "finish": "completed", "run_status": "PASS_TRANSPORT_NOT_BENCHMARK", "recorded_samples": 1, "thinking": "xhigh", "threads": 14, "cpu_moe": 48, "kv_k": "f16", "kv_v": "f16", "batch": 8192, "ubatch": 2048, "configuration_ids": [ "C096" ], "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "source_row_zero_based": 3, "details": { "phase": "after-compaction file tool call", "prompt_eval_ms": 56685.39, "eval_ms": 12367.67, "outcome": "stock apply_patch executed, native fileChange and three diff updates" } }, { "id": "M0406", "run": "functional195", "case": "after-compaction tool result continuation", "engine": "llama.cpp upstream", "commit": "8345f333951c661d166b00e6f9362e553768f292", "quantization": "UD-IQ3_XXS", "class": "functional-transport", "context_tokens": 262144, "input_tokens": 48981, "fresh_input_tokens": 99, "cached_input_tokens": 48882, "output_tokens": 69, "prefill_tps": 101.71, "decode_tps": 28.99, "quality_accepted": null, "natural_stop": true, "finish": "completed", "run_status": "PASS_TRANSPORT_NOT_BENCHMARK", "recorded_samples": 1, "thinking": "xhigh", "threads": 14, "cpu_moe": 48, "kv_k": "f16", "kv_v": "f16", "batch": 8192, "ubatch": 2048, "configuration_ids": [ "C096" ], "source": { "path": "jarvis-fixes195/native-request-timings.json", "bytes": 5024, "sha256": "8cd2b39ce8898270b97756846d956bbb4cbcf2aa04bd6e1057a1976ee9f3054b" }, "source_row_zero_based": 4, "details": { "phase": "after-compaction tool result continuation", "prompt_eval_ms": 973.34, "eval_ms": 2346.0, "outcome": "natural completed, POST_COMPACTION_OK" } } ], "voice_tests": [ { "id": "jarvis-voice182-acceptance-r2", "source": { "path": "jarvis-flash174/report/voice182/jarvis-voice182-acceptance-r2.json", "bytes": 16057, "sha256": "b3775932d5fa96a6c1e4e37a78c0ac0efcfde37d5da3511e1e8c177f8bc60e6a" }, "status": "FAIL_VOICE_ACCEPTANCE", "physical_context": 262144, "tts_full_wav_p50_seconds": null, "tts_full_wav_p95_seconds": null, "minimum_free_vram_mib": 7967, "minimum_available_ram_bytes": 37850583040, "performance": "Not a deep-context performance benchmark; prior CPU46/u4096 speed does not describe this configuration", "pending": "Listening quality, microphone, browser playback, interruption and true end-of-user-speech latency", "rows": [] }, { "id": "jarvis-voice182-acceptance-r3", "source": { "path": "jarvis-flash174/report/voice182/jarvis-voice182-acceptance-r3.json", "bytes": 48564, "sha256": "75dfb68ce6593970de9ec59e69e5b334df20255adb6e90082b3dc23ddb10cb64" }, "status": "PASS_STOCK_VOICE_TRANSPORT_NOT_MICROPHONE_ACCEPTANCE", "physical_context": 262144, "tts_full_wav_p50_seconds": 0.829, "tts_full_wav_p95_seconds": 2.41, "minimum_free_vram_mib": 4317, "minimum_available_ram_bytes": 34344992768, "performance": "Not a deep-context performance benchmark; prior CPU46/u4096 speed does not describe this configuration", "pending": "Listening quality, microphone, browser playback, interruption and true end-of-user-speech latency", "rows": [ { "index": 1, "tts_full_wav_seconds": 42.304, "audio_seconds": 3.84, "asr_seconds": 1.584, "recognized_words": 10, "asr_word_error_rate": 0.1, "llm_generating_at_tts_start": false }, { "index": 2, "tts_full_wav_seconds": 0.8, "audio_seconds": 2.96, "asr_seconds": 0.614, "recognized_words": 7, "asr_word_error_rate": 0.0, "llm_generating_at_tts_start": false }, { "index": 3, "tts_full_wav_seconds": 0.785, "audio_seconds": 3.04, "asr_seconds": 0.591, "recognized_words": 5, "asr_word_error_rate": 0.6, "llm_generating_at_tts_start": false }, { "index": 4, "tts_full_wav_seconds": 1.104, "audio_seconds": 5.28, "asr_seconds": 0.935, "recognized_words": 8, "asr_word_error_rate": 0.25, "llm_generating_at_tts_start": false }, { "index": 5, "tts_full_wav_seconds": 0.829, "audio_seconds": 3.32, "asr_seconds": 0.753, "recognized_words": 9, "asr_word_error_rate": 0.0, "llm_generating_at_tts_start": false }, { "index": 6, "tts_full_wav_seconds": 0.804, "audio_seconds": 3.12, "asr_seconds": 0.754, "recognized_words": 7, "asr_word_error_rate": 0.2857, "llm_generating_at_tts_start": false }, { "index": 7, "tts_full_wav_seconds": 0.769, "audio_seconds": 2.92, "asr_seconds": 0.509, "recognized_words": 4, "asr_word_error_rate": 1.0, "llm_generating_at_tts_start": false }, { "index": 8, "tts_full_wav_seconds": 0.859, "audio_seconds": 3.56, "asr_seconds": 0.671, "recognized_words": 7, "asr_word_error_rate": 0.375, "llm_generating_at_tts_start": false }, { "index": 9, "tts_full_wav_seconds": 0.815, "audio_seconds": 3.24, "asr_seconds": 0.726, "recognized_words": 7, "asr_word_error_rate": 0.2857, "llm_generating_at_tts_start": false }, { "index": 10, "tts_full_wav_seconds": 0.794, "audio_seconds": 3.12, "asr_seconds": 0.576, "recognized_words": 6, "asr_word_error_rate": 0.1667, "llm_generating_at_tts_start": false }, { "index": 11, "tts_full_wav_seconds": 0.823, "audio_seconds": 3.32, "asr_seconds": 0.744, "recognized_words": 7, "asr_word_error_rate": 0.1429, "llm_generating_at_tts_start": false }, { "index": 12, "tts_full_wav_seconds": 0.954, "audio_seconds": 4.24, "asr_seconds": 0.819, "recognized_words": 10, "asr_word_error_rate": 0.1, "llm_generating_at_tts_start": false }, { "index": 13, "tts_full_wav_seconds": 0.841, "audio_seconds": 3.48, "asr_seconds": 0.731, "recognized_words": 7, "asr_word_error_rate": 0.1429, "llm_generating_at_tts_start": false }, { "index": 14, "tts_full_wav_seconds": 0.693, "audio_seconds": 2.48, "asr_seconds": 0.565, "recognized_words": 4, "asr_word_error_rate": 0.0, "llm_generating_at_tts_start": false }, { "index": 15, "tts_full_wav_seconds": 0.821, "audio_seconds": 3.48, "asr_seconds": 0.655, "recognized_words": 6, "asr_word_error_rate": 0.0, "llm_generating_at_tts_start": false }, { "index": 16, "tts_full_wav_seconds": 0.808, "audio_seconds": 3.24, "asr_seconds": 0.643, "recognized_words": 7, "asr_word_error_rate": 0.0, "llm_generating_at_tts_start": false }, { "index": 17, "tts_full_wav_seconds": 1.05, "audio_seconds": 4.84, "asr_seconds": 0.939, "recognized_words": 8, "asr_word_error_rate": 0.4286, "llm_generating_at_tts_start": false }, { "index": 18, "tts_full_wav_seconds": 0.91, "audio_seconds": 3.84, "asr_seconds": 0.786, "recognized_words": 5, "asr_word_error_rate": 0.3333, "llm_generating_at_tts_start": false }, { "index": 19, "tts_full_wav_seconds": 1.299, "audio_seconds": 6.76, "asr_seconds": 1.591, "recognized_words": 17, "asr_word_error_rate": 1.4444, "llm_generating_at_tts_start": false }, { "index": 20, "tts_full_wav_seconds": 2.41, "audio_seconds": 11.2, "asr_seconds": 2.384, "recognized_words": 22, "asr_word_error_rate": 0.1818, "llm_generating_at_tts_start": false } ] }, { "id": "jarvis-voice182-acceptance-r4", "source": { "path": "jarvis-flash174/report/voice182/jarvis-voice182-acceptance-r4.json", "bytes": 5704, "sha256": "67d8dcc1b371c3b89c9441f1e269f01a7f08dd5119267c8d4f88d5b0f8ebee11" }, "status": "PASS_STOCK_VOICE_TRANSPORT_NOT_MICROPHONE_ACCEPTANCE", "physical_context": 262144, "phrase_count": 1, "voice_first_no_thinking": true, "tts_full_wav_p50_seconds": 5.877, "tts_full_wav_p95_seconds": 5.877, "minimum_free_vram_mib": 5243, "minimum_available_ram_bytes": 35580780544, "performance": "Not a deep-context performance benchmark; prior CPU46/u4096 speed does not describe this configuration", "pending": "Listening quality, microphone, browser playback, interruption and true end-of-user-speech latency", "rows": [ { "index": 1, "tts_full_wav_seconds": 5.877, "audio_seconds": 3.84, "asr_seconds": 1.872, "recognized_words": 10, "asr_word_error_rate": 0.1, "llm_generating_at_tts_start": true } ] }, { "id": "jarvis-voice182-acceptance", "source": { "path": "jarvis-flash174/report/voice182/jarvis-voice182-acceptance.json", "bytes": 1729, "sha256": "e2e60ae6b2fe1c267681e12f505c432a3a5838f707107e6efdbcc7d6d512fdb7" }, "status": "FAIL_VOICE_ACCEPTANCE", "physical_context": 262144, "tts_full_wav_p50_seconds": null, "tts_full_wav_p95_seconds": null, "minimum_free_vram_mib": 7967, "minimum_available_ram_bytes": 38003482624, "performance": "Not a deep-context performance benchmark; prior CPU46/u4096 speed does not describe this configuration", "pending": "Listening quality, microphone, browser playback, interruption and true end-of-user-speech latency", "rows": [] } ], "source_manifest": [ { "path": "context176-262144-cpu48-q8-20261005/ctx262144-cpu48-q8_0-b8192/configuration.json", "bytes": 2147, "sha256": "2d7c52b17b7a4b0a5182ea8b4189b12d74e36636a0e9f4a319195fa1fc300251" }, { "path": "context176-262144-cpu48-q8-20261005/ctx262144-cpu48-q8_0-b8192/server.log", "bytes": 19288, "sha256": "e48c08e9b4086831d4c56108804b6c3c42b67c390ecdc500ac6c6a615b0eed60" }, { "path": "context176-262144-cpu48-q8-20261005/requests.csv", "bytes": 146, "sha256": "eb9e980aac7ace44eefe900bca780d2dc89afb76f9d4037baf7df4fd0ed1161f" }, { "path": "context176-262144-cpu48-q8-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, { "path": "context176-262144-cpu48-q8-20261005/resources.jsonl", "bytes": 104399, "sha256": "be8525a2b0779fd72443ba71e994b541906fce92745836ba23b8f7651fef7002" }, { "path": "context176-262144-cpu48-q8-20261005/summary.json", "bytes": 5731, "sha256": "4ebf123dc08261f208ca6ec14c7daf11855d039e5eb086ecf8430370fb6ca666" }, { "path": "context176-262144-cpu48-q8-u1024-20261005/ctx262144-cpu48-q8_0-b8192-u1024/configuration.json", "bytes": 2147, "sha256": "e2cb8a7ff9f14b47988a588a4b56e9f69541b1fec30f6ef8bc37a24265b03139" }, { "path": "context176-262144-cpu48-q8-u1024-20261005/ctx262144-cpu48-q8_0-b8192-u1024/server.log", "bytes": 15240, "sha256": "f77b3ee628c70314825ef98718516b2f3e6922def8c83a5714f3e660e5b3cf61" }, { "path": "context176-262144-cpu48-q8-u1024-20261005/requests.csv", "bytes": 146, "sha256": "eb9e980aac7ace44eefe900bca780d2dc89afb76f9d4037baf7df4fd0ed1161f" }, { "path": "context176-262144-cpu48-q8-u1024-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, { "path": "context176-262144-cpu48-q8-u1024-20261005/resources.jsonl", "bytes": 80167, "sha256": "e8a23a721a81a2a331636257fdadfd1c4467eea2ace19fe41a6e4a04d30ed816" }, { "path": "context176-262144-cpu48-q8-u1024-20261005/summary.json", "bytes": 5737, "sha256": "76641ff146d06b18215e41dcad101da9b6ef6fb0ba143165a17e6b26e59634e7" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/configuration.json", "bytes": 2147, "sha256": "eb0739664396eaa29e7e1196ccfe94a8e203ab42d109add7f99a0280d6c4b4f6" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-131072-0/metrics-after.txt", "bytes": 2912, "sha256": "e5d107d7202673d2f5b18651716cb792d36e89ea2d1d6335f6164d5ffab10948" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-131072-0/metrics-before.txt", "bytes": 2901, "sha256": "b2fc8cbcfea79a6e4631ceb2686783e5e774fabd6fe3a37abc59805706656523" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-131072-0/output.txt", "bytes": 837, "sha256": "64096f447b6b52e992e1ededeb5f7ace1723d78a509f1fb0100e0d5cdd1d5f3f" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-131072-0/payload.json", "bytes": 1139515, "sha256": "1496901b3b74ec4ed9169c4dd38dd74cbfc2787ae8236e17b2a7dfca8e878eaa" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-131072-0/receipt.json", "bytes": 56059, "sha256": "e1b9eaf2427ed62afc41b174a0c38d9d38f3bfd214637521f30d0e335a3039f0" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-196608-0/metrics-after.txt", "bytes": 2915, "sha256": "392c2bddbb533b241089111596dce876431f093cf9b93f289feef834e63b7527" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-196608-0/metrics-before.txt", "bytes": 2902, "sha256": "0a557c7cdd7c1fb125b2d0cd2f9beb91ed0754b83b0a847ad5630722be44c96f" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-196608-0/output.txt", "bytes": 882, "sha256": "89bcde1aaebe102fa74f33680e89362c40defbb76447bd0a25c1dc5b9aae3cf2" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-196608-0/payload.json", "bytes": 1709090, "sha256": "6b1997716126f2816110c72675f94064a99d087cf7163371a6cf68dbff1efee1" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-196608-0/receipt.json", "bytes": 94472, "sha256": "a9d0f3f8e445afdce3c58efa4f1d23d34a22cc95cfa04255bd8ca3728eb514f4" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-235929-0/metrics-after.txt", "bytes": 2921, "sha256": "f3d3a3ad42d43fc420271212af3e0a2359ed87f94b7fb3fdcdafcfced83589d4" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-235929-0/metrics-before.txt", "bytes": 2908, "sha256": "f321612608765e6331c45773884c4ba2fcd902c0fd7466ba4cad7c52ddc135ed" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-235929-0/output.txt", "bytes": 655, "sha256": "982aaa34735d163cf8dc703258d19fab2e6a543f46dc1dd297153f5d9804e770" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-235929-0/payload.json", "bytes": 2050818, "sha256": "8a53e56f2ed23b00dfb1e6ac4046f9e652c55181d5d2a0156684a124eab8a32b" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-235929-0/receipt.json", "bytes": 119676, "sha256": "83510abfd404fd2a5e5a3400b42b5cda525bcd31f4e9f682b079d76dba12abe9" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-65536-0/metrics-after.txt", "bytes": 2911, "sha256": "40c4601d8ac4cabb1b4d797a62c164d52706e0266cec5d17b79008c30a30172e" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-65536-0/metrics-before.txt", "bytes": 2897, "sha256": "90d520ade0c38dd0ddd09e0b856f0c8bc1d67214857829ada652cd2974086a88" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-65536-0/output.txt", "bytes": 725, "sha256": "9902f556f65c95f3bf88121822c5fee777dd6dabbe83c2a5f478323caf3e5be3" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-65536-0/payload.json", "bytes": 569952, "sha256": "564a65d950c2f6968de74b59f43b86a86262a21904c088fee16e9edc469dae86" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/quality-65536-0/receipt.json", "bytes": 26594, "sha256": "77fbf0ff4db4f11c9dd73f3ad0a67c85dbd36217398e0a9b94aa3b3cb40c7a0e" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/server.json", "bytes": 356, "sha256": "6d51b3741bd0b0fb32d4904dbd8478523f85b077e63674f58694c246236fc312" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/server.log", "bytes": 121771, "sha256": "28e9d68d939bc18f0369163e1a7bb946ea6448980eb52415c8b5226d189454fa" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-131072-0/metrics-after.txt", "bytes": 2913, "sha256": "6ed2ad5771807afaea877fbae585d0ea6692213117d1d80d619c24cbd5e25f5f" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-131072-0/metrics-before.txt", "bytes": 2899, "sha256": "3ede0809cadaa1a0c6c3a3fb995750decd530b6d0db1bf47b64e5427c73806ee" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-131072-0/output.txt", "bytes": 1347, "sha256": "238f0d30269380bb98fb45031d2e0d0c6080d943c4dbf80bf64e796ab34d4f72" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-131072-0/payload.json", "bytes": 1139533, "sha256": "2bdfeeb8e20c755b8f214b40a84f3d927ec51ad3fca6b28b25cefd163029ac5e" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-131072-0/receipt.json", "bytes": 58125, "sha256": "45ae9e0bf9504d7d663d2b05262095d305d9dca081bbef4560712b5121ada8a9" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-196608-0/metrics-after.txt", "bytes": 2914, "sha256": "2c7c707ef1761df3c867eac0ff8ae16e6a2796a93b48b3c09b264b6122e33714" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-196608-0/metrics-before.txt", "bytes": 2901, "sha256": "e5b4b1556fabfc8adee864024e15bee384c61dfe8c2590e16e6e778c0c4bd556" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-196608-0/output.txt", "bytes": 1381, "sha256": "f4f84ba5c92810c17598a0e1a6f50fe8c10b309ae58b8b6872005a7830cb5ba7" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-196608-0/payload.json", "bytes": 1709111, "sha256": "4e012c6916975dbe85751a766aa8d0359ff3bc52bc9e257706bba4cac012e9a0" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-196608-0/receipt.json", "bytes": 96568, "sha256": "6a2e7df947d4023aaf6f35540d9a31a701813e42cd95b7d09317c1f86740e3c5" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-235929-0/metrics-after.txt", "bytes": 2920, "sha256": "44a3960c87a864fcd4acca9a4641998313898687499913fdf275e95ebbc41bf5" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-235929-0/metrics-before.txt", "bytes": 2903, "sha256": "41260bafb72c42d08c5880ce5e2ff6ab4e0c97eaef4b4a57f9fa29410a9f2b0d" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-235929-0/output.txt", "bytes": 1660, "sha256": "70a515ab8872370f2c9360e457155acda293240213481411c0883ee3e6ca538c" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-235929-0/payload.json", "bytes": 2050852, "sha256": "e45c84a6cd216fd45768d8b83210d2c270dc3274a79884f13f005a9dfc592fb0" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-235929-0/receipt.json", "bytes": 123322, "sha256": "c03b131d95e9808850f388b03920b26270d82590cad08ab99c55e6e54e33729c" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-65536-0/metrics-after.txt", "bytes": 2909, "sha256": "72584965ee182571ebe661796bc1951ee46a0a1bfe89a363ecaec89b33a0bad7" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-65536-0/metrics-before.txt", "bytes": 2894, "sha256": "757d122f2151eab61d05ef9f3a6c044e71807ddf3febbddb43457f953f53ebc1" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-65536-0/output.txt", "bytes": 1340, "sha256": "3f6b1144dd98fdaa3cd536917bdb6e4a790b241d29ae361e872663eae4d630cb" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-65536-0/payload.json", "bytes": 569957, "sha256": "a4404fbfffecf74439a72dc09e4c6c740d8f6914b4b9f12b3abf8373e55d5eb5" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-65536-0/receipt.json", "bytes": 28317, "sha256": "d7e5f44ab5877dcb3a2284622ce47997c292b71bcacc9b32f9ccb02b8172767e" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-8192-0/metrics-after.txt", "bytes": 2906, "sha256": "d22162effd8c50dad00602cce8d14d562507eda3ac52de0f386d4d436362f64e" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-8192-0/output.txt", "bytes": 1260, "sha256": "52b8a3d281f5eb31ad7bac78380de3c297b39be192a521540a39e015af92184a" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/ctx262144-cpu48-q8_0-b8192-u2048/speed-8192-0/receipt.json", "bytes": 9691, "sha256": "16c905e8b7de57a0fad7ff18163ed17bb738ad786d513553b471e3cb5507b6eb" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.csv", "bytes": 1530, "sha256": "81ac471ae3717518e9db606a3ccddb38881c1b02b451b1f60903d0630f36abfc" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/requests.json", "bytes": 659472, "sha256": "d817ddb499326febee1d24d82d1a7316db16fb40639b17f6bc6ff2d9c71b9bb2" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/resources.jsonl", "bytes": 4159718, "sha256": "7602292a088d4d15d7023c38cab6773110e47a158230eb89bd85c1a49dbdc815" }, { "path": "context176-262144-cpu48-q8-u2048-20261005/summary.json", "bytes": 7186, "sha256": "7861713386491472033365b7ed17d46b12127c87fe29bd94cc32257aa9ad2173" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/configuration.json", "bytes": 2146, "sha256": "bfbc368e38b5ef2bb4a85a4d66565236e05d7dedcfc41fce73c27f1a03e0d771" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-16384-0/metrics-after.txt", "bytes": 2914, "sha256": "53ff634c23fe58c5d9475e1b7c72fea0b8518c713067fcdd8fe2ff431df11f83" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-16384-0/metrics-before.txt", "bytes": 2902, "sha256": "a1546e2fe4985e8746096303a9a07bc32df647d14fc128e0652e4ede76c9986e" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-16384-0/output.txt", "bytes": 590, "sha256": "eaaded4ad776a5775cfe44335091e024bca64dc6391c1f132f8c74c5483cba03" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-16384-0/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-16384-0/receipt.json", "bytes": 7963, "sha256": "7cc3222579175d6cfc82444244f5c8751cc3a51772b2465a2662677376230ddf" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-24576-0/metrics-after.txt", "bytes": 2916, "sha256": "220cd7226e784618ba72451e3caf21f1eb18e39902cf120e7a1ffa6b9badd1aa" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-24576-0/metrics-before.txt", "bytes": 2904, "sha256": "eb0e020c754747ca49c2f0adde4e1c50757eaf9b56fff0920cdba42bfcccc457" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-24576-0/output.txt", "bytes": 623, "sha256": "67aacedf18649a2cc3b51e8c84ad5d83ae0d9e533c7274e4a98dc4c7764b14b8" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-24576-0/payload.json", "bytes": 213944, "sha256": "bbd9d2060737d50ef1a5ed82775116d2070277de7396b482f689d176c18057ba" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-24576-0/receipt.json", "bytes": 9844, "sha256": "d74aa3a87d3c7621e18a517595a037b14b2658c94b8598b152228586704c1690" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-29491-0/metrics-after.txt", "bytes": 2914, "sha256": "703db3c27f90d96e7412592cbef0f83a7204d3b3c78f803b8f0090293d3a9bc3" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-29491-0/metrics-before.txt", "bytes": 2904, "sha256": "76b93533d519651310dfc5cb40bf7132686bb809de095e0c0aa98e5e1d74a7bb" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-29491-0/output.txt", "bytes": 626, "sha256": "0ac54873f01cb1e56542923a77612724dbd6c6c4cafa11ee30ed8ef3f2c0070e" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-29491-0/payload.json", "bytes": 256653, "sha256": "cef65d2a36d064ed8c349ea1891d05ed97129c141617350d8ee3f7c637a64e03" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-29491-0/receipt.json", "bytes": 10682, "sha256": "79a82cb7b01616f7b577a863d4c0c1ad070bef0b1b95b182e7219cc7c23e75a9" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-8192-0/metrics-after.txt", "bytes": 2908, "sha256": "5b734082dd22916c03fe67fb7e2415b8c61e559f6d5abcbf08b1ebeee7ab31a5" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-8192-0/metrics-before.txt", "bytes": 2897, "sha256": "963e8978d406addff846050058329e1287146cadea034e37f5f74c339f1c3319" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-8192-0/output.txt", "bytes": 559, "sha256": "bcc9b36fb5343f2e93238c2a5ed3b9cb844eff09ca703a8e7276bd417f3dd654" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-8192-0/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/quality-8192-0/receipt.json", "bytes": 6791, "sha256": "9ae0f441720b775eabb2aab3e3cc7c323eef34c6d78ca4bd3770c97b4fdab444" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/server.log", "bytes": 114944, "sha256": "fe7d5961ec1fb5461e08e945dd423ecf09f46764a0280b6f72c0523fc1f61c09" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-0/metrics-after.txt", "bytes": 2911, "sha256": "80f36bbe353583c6a678955e5c302640b26c4e31c8ed5689710fb20178568846" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-0/metrics-before.txt", "bytes": 2896, "sha256": "b924d68add35cd2a908185cfe7bec3def17730908fcc38f4c56b4acad4b8e5f5" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-0/output.txt", "bytes": 1252, "sha256": "d11ac8ace3bd59ccad7e4db2c8c672e9ecde3c451706ac9e8782241803d2b1e1" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-0/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-0/receipt.json", "bytes": 9654, "sha256": "29da5b9a95a14abc139e51901b738d09f116483bbe88e55ee2ad9ad5e6579bd6" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-1/metrics-after.txt", "bytes": 2912, "sha256": "69db1e5a14238d2b68c55456be3f3a3b3d6f7768a2486797e31ec1c5ce2fe10f" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-1/metrics-before.txt", "bytes": 2899, "sha256": "821f6107fc2ad9f23b43a83870fa02191d5c46edf195e0d13d4cc2dfdc65f347" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-1/output.txt", "bytes": 1252, "sha256": "d11ac8ace3bd59ccad7e4db2c8c672e9ecde3c451706ac9e8782241803d2b1e1" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-1/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-1/receipt.json", "bytes": 9656, "sha256": "096eeef382ac2d42c5eedf3e9262ba81c55f7c1079299d8ca5cc77de5a676254" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-2/metrics-after.txt", "bytes": 2914, "sha256": "7a4b234a4a4a06aadc33ada3fde574f1875612ba01acc6acc009b4296f3112ee" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-2/metrics-before.txt", "bytes": 2900, "sha256": "c9f7636effc439f930b69510cd81785489fb6d56995761b6e93395b1a2dbf5b5" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-2/output.txt", "bytes": 1252, "sha256": "d11ac8ace3bd59ccad7e4db2c8c672e9ecde3c451706ac9e8782241803d2b1e1" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-2/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-16384-2/receipt.json", "bytes": 9659, "sha256": "3ae5d7d6ffa201c7ada12bc363fa18a4393506248340655fdf5306e901ab5107" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-0/metrics-after.txt", "bytes": 2916, "sha256": "157b33659f02eca21a65cc5724c22691710f073741aef1a77bfd9c1201ecff5b" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-0/metrics-before.txt", "bytes": 2902, "sha256": "118829abafdfc160be4f43ddf32d59b3028bbf4ccf4c202c8a0b7d1d39efaf38" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-0/output.txt", "bytes": 1509, "sha256": "60c44a35d6b8099603bba276968eeac480086719d5b1d580783b2761ebb8d566" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-0/payload.json", "bytes": 213976, "sha256": "2b41781ddc2a0c2ba2285c74398d20154cc9a6695b39adf3b9f8b2b6597229fa" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-0/receipt.json", "bytes": 11589, "sha256": "651979337d0ffdfd1fd28c3220e284bd69e49cdb2a74e8473a964de5991550d1" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-1/metrics-after.txt", "bytes": 2916, "sha256": "c765b9cf2865bbf71ac5f4ec6a9d4a740afa2b01355c356c0e8501eaa91c98c9" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-1/metrics-before.txt", "bytes": 2904, "sha256": "2a78cad5a38a92afec4199c120a5a32992e9f18043f9d8fe14010e06f606e510" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-1/output.txt", "bytes": 1509, "sha256": "60c44a35d6b8099603bba276968eeac480086719d5b1d580783b2761ebb8d566" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-1/payload.json", "bytes": 213976, "sha256": "2b41781ddc2a0c2ba2285c74398d20154cc9a6695b39adf3b9f8b2b6597229fa" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-1/receipt.json", "bytes": 11204, "sha256": "610873291763e86d6983497455e18418ca0f0486efa8f7c976546f208a878493" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-2/metrics-after.txt", "bytes": 2916, "sha256": "1ffe47a9149bcf1285bd3c143674646d62e08e5486be0fd82171e99b0aed2671" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-2/metrics-before.txt", "bytes": 2904, "sha256": "8ed27b7aaa464a9b5312024ea88024eb1603183005c17b5d7382ad1c066be1f2" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-2/output.txt", "bytes": 1509, "sha256": "60c44a35d6b8099603bba276968eeac480086719d5b1d580783b2761ebb8d566" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-2/payload.json", "bytes": 213976, "sha256": "2b41781ddc2a0c2ba2285c74398d20154cc9a6695b39adf3b9f8b2b6597229fa" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-24576-2/receipt.json", "bytes": 11197, "sha256": "6222fd9ea9f1d6204cbe56573a08d8ad905559147d046bfc7413968b5090c766" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-0/metrics-after.txt", "bytes": 2916, "sha256": "4928e6d10dafb9e49926b93ede30fda2d933edac3572e2805a85215d4fad4f97" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-0/metrics-before.txt", "bytes": 2904, "sha256": "d2e90d32f05f96ac72147492526b5a9450f06ed0fb47c45f8897ec5ad088f467" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-0/output.txt", "bytes": 1397, "sha256": "e5e577ebbff2567cd6fa63fa4a8abcecc006d74e40d36218f53778335ec8bc3b" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-0/payload.json", "bytes": 256685, "sha256": "1fc27c824a011827f000b33d9eeab0e72957e5304d62f30db9ff1141772510b5" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-0/receipt.json", "bytes": 12378, "sha256": "9ae6dff43c193b7ded46be36b062a4d5496ba75de9de9b6b9417cc22d5fc7041" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-1/metrics-after.txt", "bytes": 2916, "sha256": "2c296b6f66f4c04dec82f0ce22686894838c179ccb71d8c54c68f2b586daf47a" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-1/metrics-before.txt", "bytes": 2904, "sha256": "740ea39b54b8e59db958c4275d000ac3b0c9e6e104abceb4a74b86fc6c923647" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-1/output.txt", "bytes": 1397, "sha256": "e5e577ebbff2567cd6fa63fa4a8abcecc006d74e40d36218f53778335ec8bc3b" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-1/payload.json", "bytes": 256685, "sha256": "1fc27c824a011827f000b33d9eeab0e72957e5304d62f30db9ff1141772510b5" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-1/receipt.json", "bytes": 12381, "sha256": "70b804dfc16f9949199bade96f2c5f7ba69d654f49668db7fe23200a4ace1522" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-2/metrics-after.txt", "bytes": 2916, "sha256": "852a2d616b50a7911c575756932d0ddb5b57bbcea01a2b70e874d9cb5296fff9" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-2/metrics-before.txt", "bytes": 2904, "sha256": "01cc75f9454ef06a3c537ae3c880a7e7967c25a7796fb22644ea37bcb0a230ac" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-2/output.txt", "bytes": 1397, "sha256": "e5e577ebbff2567cd6fa63fa4a8abcecc006d74e40d36218f53778335ec8bc3b" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-2/payload.json", "bytes": 256685, "sha256": "1fc27c824a011827f000b33d9eeab0e72957e5304d62f30db9ff1141772510b5" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-29491-2/receipt.json", "bytes": 12372, "sha256": "6631fd50e00b6d54b10cef5e95953ff7a5a2937ad574e7cb428e3b1228c8fc1d" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-0/metrics-after.txt", "bytes": 2906, "sha256": "15f6dddd71a77b3aa413ae0afa83618bcff0fb73612213fc0119d2aea7d1f315" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-0/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-0/receipt.json", "bytes": 8481, "sha256": "13c692217c78be0badf77cbfecdf7a1733a5caa095e3a3a076bbe3ee99de09a6" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-1/metrics-after.txt", "bytes": 2907, "sha256": "ee67311b6a9ba47b1237d8e149d9e09893eb826df45dbcfa0ee68f8e4ccd4443" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-1/metrics-before.txt", "bytes": 2894, "sha256": "9631636467e1e3b266dd524d36c4f4f00a4546296360844f8e21cceeef10a7b6" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-1/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-1/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-1/receipt.json", "bytes": 8482, "sha256": "c87ea85060e12ae27835dc2b3df950ba45efc4186aeb96639763322fe39aef35" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-2/metrics-after.txt", "bytes": 2909, "sha256": "b0f0d9ce67ab377f964882e0211af98fb0e772af90a579ece1939e2de30ee248" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-2/metrics-before.txt", "bytes": 2895, "sha256": "9dd42791a3f1b6bd6637c8928c7fe05c52dec279599294c569384361809d768f" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-2/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-2/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu43-q8-20261005/ctx32768-cpu43-q8_0-b8192/speed-8192-2/receipt.json", "bytes": 8483, "sha256": "0ce61f87dd0483e818881438be369c7e5b95257bd5dcce8012de4151d44f40eb" }, { "path": "context176-32768-cpu43-q8-20261005/requests.csv", "bytes": 2394, "sha256": "7e0da760047dbbec93ee1cf979c238c48dcf9096d09052aa7388ead0350f085f" }, { "path": "context176-32768-cpu43-q8-20261005/requests.json", "bytes": 171903, "sha256": "ffea839be2fc1d086fbea5251cc01cb588b0716dfbb50733eea00a22f0c3b5f4" }, { "path": "context176-32768-cpu43-q8-20261005/resources.jsonl", "bytes": 786312, "sha256": "f9bc5a61628af417282555c9436de5fa1cd5e16631a1dcdd985524699ad4c58e" }, { "path": "context176-32768-cpu43-q8-20261005/summary.json", "bytes": 6866, "sha256": "7fcbe1ff12affeb72174e9b745ea11909e14b4860c88e6b5d2bc339ab793d7c2" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/configuration.json", "bytes": 2146, "sha256": "5c7c5471de304f68ca306df546ec5590a1ed9ef63bd3e493a50d35ab8219d045" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/quality-8192-0/metrics-after.txt", "bytes": 2910, "sha256": "0e73d279ddc5c5d95ab6d7fc1fa5d99839c213d991e8a56fd2db54a0e4a1ec56" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/quality-8192-0/metrics-before.txt", "bytes": 2897, "sha256": "fef0a02bfdd6ab980153143d50c31d793a515bc479791e5084f6aad9057d7f1f" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/quality-8192-0/output.txt", "bytes": 564, "sha256": "239a4cff169091eba6e4e45546a14d2134b5f53f03e5bf81fe85ec6628957c29" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/quality-8192-0/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/quality-8192-0/receipt.json", "bytes": 6801, "sha256": "bbebe2f638620a23a3ef27e5e150b650cbf79ab6bd3110182a71560377170c85" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/server.log", "bytes": 53835, "sha256": "7f89399bda773648f2666b7cc62deec2023ac2892be45cd9df1339e3d132fd5e" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-0/metrics-after.txt", "bytes": 2906, "sha256": "4b18b1cf972163759f139daff213bd40483228b5e7b07955fab8ceaa66a86cc6" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-0/output.txt", "bytes": 1342, "sha256": "8acc7781e3ebb89d084754749f6d63e26dbe630b87fc6060df9a89fce9bf8b18" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-0/receipt.json", "bytes": 8875, "sha256": "dd338e929902534e6923144c5ac08352ec1a580241bbceb54a0f83ff646e6fa6" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-1/metrics-after.txt", "bytes": 2908, "sha256": "bd7fca7a5250698a029e08140ab7261eb434466bdfed1d4b3c51d66892056729" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-1/metrics-before.txt", "bytes": 2894, "sha256": "a4f1afd09965d06389d0444e8fff94c1a37688c626f4fef2872a2b57dbc53d2d" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-1/output.txt", "bytes": 1342, "sha256": "8acc7781e3ebb89d084754749f6d63e26dbe630b87fc6060df9a89fce9bf8b18" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-1/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-1/receipt.json", "bytes": 8496, "sha256": "2376e4c85e3781ca10caccc13b85ab1c7de5ffbbf30d0acf6125ddc6e8793587" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-2/metrics-after.txt", "bytes": 2909, "sha256": "6920791111d3b8228a953fbe491e3e9a38c807360d9d8a4380861268eda91e2c" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-2/metrics-before.txt", "bytes": 2896, "sha256": "085f5adefbffc38fc9c0dd62fd3fea6f4b64288b2eef0f174144e2bc1559ac51" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-2/output.txt", "bytes": 1342, "sha256": "8acc7781e3ebb89d084754749f6d63e26dbe630b87fc6060df9a89fce9bf8b18" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-2/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-control-20261005/ctx32768-cpu48-q8_0-b8192/speed-8192-2/receipt.json", "bytes": 8489, "sha256": "3a2614e348d7b22b26695680d4c6a6cbf93162c86f307407b6092d376e4e0c6d" }, { "path": "context176-32768-cpu48-q8-control-20261005/requests.csv", "bytes": 698, "sha256": "270022835944a36b23a081a7d563d7861f1173ea0a04ec882aeb7a790050df05" }, { "path": "context176-32768-cpu48-q8-control-20261005/requests.json", "bytes": 34850, "sha256": "ff498936a7091a684a8353a86fd943613ec9e70fa85f1e8ba5db3637f7977a56" }, { "path": "context176-32768-cpu48-q8-control-20261005/resources.jsonl", "bytes": 231060, "sha256": "87d8506f43c15fbea577fb2fcaed2a7d207dd93d13d57ed1488ea752591e2353" }, { "path": "context176-32768-cpu48-q8-control-20261005/summary.json", "bytes": 6036, "sha256": "2aebac4f7f51cb2bcd4f6798beb53b8d93617f8bb7e1c0cb7c65fb39acd48546" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/configuration.json", "bytes": 2146, "sha256": "43566c28a2aad347a8e5366fe4772e14c128dbff6ee99de7a6eff188162e357c" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/quality-8192-0/metrics-after.txt", "bytes": 2910, "sha256": "ed1d6eb0725d26c6eace291acf4e365d76bd5dad9510ef020927faec3157c31b" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/quality-8192-0/metrics-before.txt", "bytes": 2897, "sha256": "682611e49706c22fa2676bc8694dacef82625a16bd957f9bb76fdc5bc37101cd" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/quality-8192-0/output.txt", "bytes": 594, "sha256": "5952deadfcee38d882991a491a398d30450380c46694e53550edc145f4a5f789" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/quality-8192-0/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/quality-8192-0/receipt.json", "bytes": 8770, "sha256": "e80384d032d7931bbd94ad6775ce41aa9c93246e9f4bdf4d3fab2f284f30d08c" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/server.log", "bytes": 56019, "sha256": "549c54dec1d5ace54444d5412398293f798481d6332347156ace5df59b58b4be" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-0/metrics-after.txt", "bytes": 2905, "sha256": "11958a8fe116a4a29e807b19024fb45731e9909e2ea7ca8afbb2a44e5da6ce4e" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-0/output.txt", "bytes": 1387, "sha256": "17b5c5f168aea9a2402596415ab87073197a2a6ed33c28a1b07cfc5995c6fcfe" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-0/receipt.json", "bytes": 10826, "sha256": "b40f2c93b340efd96f71e023f2052d29688828508f191febc0e2d7d02b768713" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-1/metrics-after.txt", "bytes": 2908, "sha256": "d5684750d79e86f3346d07086071e8125c54198364767667a98cc4eb18ec9553" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-1/metrics-before.txt", "bytes": 2893, "sha256": "5352c4f5540c4a5e6229e08dae5c9f26e1e319a51810f5d6c236a6aea556b40f" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-1/output.txt", "bytes": 1387, "sha256": "17b5c5f168aea9a2402596415ab87073197a2a6ed33c28a1b07cfc5995c6fcfe" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-1/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-1/receipt.json", "bytes": 10062, "sha256": "baceb58b6e0f081673b4e7670a0728367082126111685545e336a82adad63ba4" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-2/metrics-after.txt", "bytes": 2909, "sha256": "c9dda68edbb8377a3b40cf191b622ad365f4428609d18b319021b1f464a4876a" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-2/metrics-before.txt", "bytes": 2896, "sha256": "36f3b25a6941b3a023975bfed0a3b057b27855a7e0473af769b3d796dedf9522" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-2/output.txt", "bytes": 1387, "sha256": "17b5c5f168aea9a2402596415ab87073197a2a6ed33c28a1b07cfc5995c6fcfe" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-2/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/ctx32768-cpu48-q8_0-b8192-u1024/speed-8192-2/receipt.json", "bytes": 10065, "sha256": "0440b18b9de82a5b4b4fadb3a04ec6b6ac3177df77cff20ddc4ae8ff66b21bd3" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.csv", "bytes": 719, "sha256": "247025a319315ef65c1e533705a503a7a13e94e2e428404f47d009cba2c28b4e" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/requests.json", "bytes": 42460, "sha256": "d0f8d9927daefdc84aee21bf558e7c953254a9a823ba1fc860270028e4019d02" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/resources.jsonl", "bytes": 286322, "sha256": "6d249629f93fcca36e79928ca8021ec6ffa0130638d22d3ba1aa3b37d3471734" }, { "path": "context176-32768-cpu48-q8-u1024-control-20261005/summary.json", "bytes": 6069, "sha256": "efca0c31366fff671e7ad9c9bec7370227d3710ee6008fcc87106d7630ccfddb" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/configuration.json", "bytes": 2146, "sha256": "480938ce5744ef74cc07f20bae6df84fa65c0a178e94f25ae882b4ee12f7c611" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/quality-8192-0/metrics-after.txt", "bytes": 2907, "sha256": "b58b825a89870b7900c381d2c01a4e6afe37e1a9ab55f1187dc5559adc805bd9" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/quality-8192-0/metrics-before.txt", "bytes": 2893, "sha256": "df4dd6bb60685d7b4c5ebcd7baf331bfe61c4ad8bd71d66f8dee7b3db0eb5060" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/quality-8192-0/output.txt", "bytes": 639, "sha256": "e3d4e96787435087634d3659d835c0c43c4510668228560d49843d8a3c90330e" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/quality-8192-0/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/quality-8192-0/receipt.json", "bytes": 7613, "sha256": "d764bc5fccd087ee0c2967bd156c913ef94a710d38e17365973606f98aaeaf2a" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/server.log", "bytes": 46985, "sha256": "7c183f20b0a6ed6b35b3977b6637646ddfa43e4825a2d129f543db40d4990b84" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/speed-8192-0/metrics-after.txt", "bytes": 2905, "sha256": "d311d30f9b7e5b9246d656a8dc75fe038d1d2ae0da82acd77bebc777fcf7d742" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/speed-8192-0/output.txt", "bytes": 1260, "sha256": "52b8a3d281f5eb31ad7bac78380de3c297b39be192a521540a39e015af92184a" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/ctx32768-cpu48-q8_0-b8192-u2048/speed-8192-0/receipt.json", "bytes": 9303, "sha256": "9637acec7d929d09a2b99f165ef779a9811457d27cff9cdbd20eeae461b7af8a" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/requests.csv", "bytes": 442, "sha256": "a9f6452a5597db6fc9431076abaf2a76c8745cc4f4c835b3146d9ab4cecfcac3" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/requests.json", "bytes": 18057, "sha256": "466962bb262f626bd1930f939d96a9dd005892e3a93d77f30ff95dec62b558c0" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/resources.jsonl", "bytes": 173186, "sha256": "959eee2dc85952054c439503c5749cf4f3a9a2681f67384f32a642a5109c360a" }, { "path": "context176-32768-cpu48-q8-u2048-control-20261005/summary.json", "bytes": 6069, "sha256": "4a745010e05f6fffe80575588f4c49c6c5275af0d019485a4e64c89f3c5fe477" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "491e906764b52c775796c720a618cbe18d53c4b0c9522f7732eee01d1a956a24" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/quality-36864-0/metrics-after.txt", "bytes": 2909, "sha256": "81910366d7a22fa1dc7f32bbb74a458315a9b54f29797bba6766a54460e89d46" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/quality-36864-0/metrics-before.txt", "bytes": 2897, "sha256": "c37aa73876cacd3b11d2dd4e936621aaeac2ffde4936ffb1a6d34b1b7b3bfca2" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/quality-36864-0/output.txt", "bytes": 742, "sha256": "8231b0235ecd7a9d4a9b0921663d46a02975e4feae8db92d1858bf8e35cc9661" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/quality-36864-0/payload.json", "bytes": 320736, "sha256": "c21a4adca78ed601beb9b38e366adea41480da3f15d11a10e78cf34822ff5187" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/quality-36864-0/receipt.json", "bytes": 12647, "sha256": "530bf3381bad2ee95147ee143cea52a91a49ae2aa1f98c6039305a3e4d19a0fc" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/server.json", "bytes": 355, "sha256": "23b8de3bd47d063fe9dd679d0c6cdc074341cb4ec4ca8f8bf14096933271523f" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/server.log", "bytes": 52576, "sha256": "ed18267fd01eb185d94274766f4d59512e37798edfcd56a7050609b7f59a1ed3" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-36864-0/metrics-after.txt", "bytes": 2909, "sha256": "39fe60d54a49e2a6dbd842a9ef89e99b9bdddeba87204a2f681fb21d7213b83d" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-36864-0/metrics-before.txt", "bytes": 2894, "sha256": "8c288e6ec51d32d9686343b4e7007224e1e3e862168cf14785690f1a2812cf25" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-36864-0/output.txt", "bytes": 1664, "sha256": "f9ea32b2c4f8afe16dbe05d23b121f4b547253f4111dcd081145afb5ac3345dd" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-36864-0/payload.json", "bytes": 320775, "sha256": "4b46a26c49bb09de6d9d2a2dc0f61c0a14c7b13bd9f80673fdf266e3741550ce" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-36864-0/receipt.json", "bytes": 14338, "sha256": "67b1099401f7093aaab3db4d6284d140540074703315ecf517b00c133e5bb8a0" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-8192-0/metrics-after.txt", "bytes": 2906, "sha256": "f77784b5af7c7db4181048282b99e4bc4fdc08a9f5690c1338d25e8ed40ac039" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-8192-0/output.txt", "bytes": 1263, "sha256": "662786734e9822e0e951169cd03354df504e176f8b963f7a826d22f775194151" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/ctx40960-cpu44-f16-b8192-u8192/speed-8192-0/receipt.json", "bytes": 8530, "sha256": "4a22da0cd818d26efc6b9d37514167e1b149c75bcfb2c0215daefef1f0668f26" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/requests.csv", "bytes": 575, "sha256": "78e9fad69e0b4ed254f1a929b217835d0778d41d6081361db4e964c70f4e1c1a" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/requests.json", "bytes": 38009, "sha256": "19a64e38710db82a54606b61e0ab7264ed9d57f6e11337b5bb20873e2b9edf41" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/resources.jsonl", "bytes": 279569, "sha256": "002d84a8409c79efdb1ae6a8194d65b8b2f584fc0fb93efbdbe0218d19fde389" }, { "path": "context176-40960-cpu44-f16-u8192-screen-20261005/summary.json", "bytes": 6334, "sha256": "91386b2a8101d0f7891f838b03169a1a811950857f55baf4d54a0848a4fb42e8" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "1b5c0f89909e03f461d8901faa280f36c78c6c5d5bae7d880e652ea3a99259a3" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/quality-40550-0/metrics-after.txt", "bytes": 2909, "sha256": "5ff8fea78dd689561743bc0f7a1fd25f3154d4e745a4476c38de6bc5e0b4655c" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/quality-40550-0/metrics-before.txt", "bytes": 2897, "sha256": "315c46c7fa87ef5abfae162e46c180540337b1f2905c1796670f265e2da66439" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/quality-40550-0/output.txt", "bytes": 710, "sha256": "2d5e9b7f6700a964c4983265c1d7bb699f456b4cea971a48bb3b76c7f738f103" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/quality-40550-0/payload.json", "bytes": 352766, "sha256": "b717e98ae46575db0525578a6be02950a764bfd71358daff7847a8aeb96496b2" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/quality-40550-0/receipt.json", "bytes": 13441, "sha256": "041c66f06c8149be039a9ddd2906e826bd4bad5a549cb190b5722543f09f5cbd" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/server.json", "bytes": 355, "sha256": "b3b1b5fc6016edbc74ebb2c642f17fedaa0a14d639cf44a0f5612cff87783330" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/server.log", "bytes": 52577, "sha256": "58ade1ac912b57320a83a5e55ecdade7603fd60df4f6077b912e3d2540d6e180" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-40550-0/metrics-after.txt", "bytes": 2908, "sha256": "9382d23e1553dc145ff73faa21de1072e77f50242c411cfd7e1363c12478bbf7" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-40550-0/metrics-before.txt", "bytes": 2894, "sha256": "baadde62ef73312764d883de7530c0c82cc4edc417bdedb9c33a087768caf037" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-40550-0/output.txt", "bytes": 1434, "sha256": "6a9a8437910b79832a9bad9d26e48e052de535480a3ea881bf741afd55f03886" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-40550-0/payload.json", "bytes": 352811, "sha256": "a0d0a0703d11162910b9f67fb5540b8b5120afb7a39c1b55e3e5910cbca682cc" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-40550-0/receipt.json", "bytes": 14742, "sha256": "9359c64ec370f455f85a1188c718d050148e4095ea65f9881055f5a205fbe826" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-8192-0/metrics-after.txt", "bytes": 2905, "sha256": "44b6855a9d34b79111ae4c5eba5701b07bf1ba48761c2af2bec8cdcde3926301" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-8192-0/output.txt", "bytes": 1271, "sha256": "b1e9959b4635ff06f191b35eda7f8c0673a0666fe34f49d4b31a6e51e0c8ac9c" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/ctx45056-cpu45-f16-b8192-u8192/speed-8192-0/receipt.json", "bytes": 8538, "sha256": "d9d78ce23816e5ffd83e712fb2aa847c114be536ff53e8021f997ccd80d98d3b" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/requests.csv", "bytes": 617, "sha256": "1c77fd55d3a7523564c22e1c9efd04f065c53131ace20064190fe1a8aa50fc96" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 39305, "sha256": "7eeab3be497b5af10eed75e335d98028459a1845c16df85606ce2d4635989a12" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/resources.jsonl", "bytes": 284970, "sha256": "d25258226cff4a27849728c839c1be172c8e6c345913ff5846dad7343593702e" }, { "path": "context176-45056-cpu45-f16-u8192-screen-20261005/summary.json", "bytes": 6332, "sha256": "c6239aa04329a4c8aa48d6348941effda034330d2e6af44f91c00b4211ae4fff" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "9b8f25602bf689fd026f00bffb2ce94dd0e09c61987c42833db4af0b7ff02e68" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/quality-42393-0/metrics-after.txt", "bytes": 2910, "sha256": "02b6755d49097dc2ac59c73c66ee7c277be127d1656a7c2c93822d474286c676" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/quality-42393-0/metrics-before.txt", "bytes": 2896, "sha256": "021b76807323ab748c0f700e79de2c606b67bd18a0725333e26b968acebd2019" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/quality-42393-0/output.txt", "bytes": 894, "sha256": "a43929a78836cadb027df3b08290159d0793e13831825a84d68d726594629622" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/quality-42393-0/payload.json", "bytes": 368788, "sha256": "3a5aa20823505a55501c49e22384a249956a795c846d4fcf4eacfb6a298dd013" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/quality-42393-0/receipt.json", "bytes": 14217, "sha256": "2b63e62a2aaaecb3d9aded32fab7742398493469d04f247bac5df9b3a536a5b7" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/server.json", "bytes": 355, "sha256": "989f043cbb62f0c8819b038580cc9eef9edb6f6cec4fe7cc12b2ae84c317e88f" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/server.log", "bytes": 53084, "sha256": "22641526c725b5135893b11c60b11cddbdd4c0d7124f9001291a29475edf2677" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-42393-0/metrics-after.txt", "bytes": 2908, "sha256": "8487838e459808af0e2888082f97e8581ed12a6cb2cd439e62545d2f411d8842" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-42393-0/metrics-before.txt", "bytes": 2894, "sha256": "56da94e76caf1ca5657cd431116f2c52787a7822a5fb4fc6e722010cf0606576" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-42393-0/output.txt", "bytes": 1616, "sha256": "b49b5d0086647bdaca3dc0274f93aad70648bfad6f506f3af2d40ffec69b9a41" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-42393-0/payload.json", "bytes": 368823, "sha256": "a80e3ef2830c83d1491b6e42d0befde00b0118489afe4a003572774bc18639d8" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-42393-0/receipt.json", "bytes": 15902, "sha256": "59c1ef438792944b4f3f39996031def8b3d7d727884a9064092857136d3ff489" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-8192-0/metrics-after.txt", "bytes": 2906, "sha256": "6f26dfe4138bb335ae6c9c865e2d0e8d3231e2923401c27e56630b1ffcb9cfa6" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-8192-0/output.txt", "bytes": 1271, "sha256": "b1e9959b4635ff06f191b35eda7f8c0673a0666fe34f49d4b31a6e51e0c8ac9c" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/ctx47104-cpu45-f16-b8192-u8192/speed-8192-0/receipt.json", "bytes": 8539, "sha256": "bdc2564c8ec0512e383c065d1718ca3451baeecf41f381f5d611706943a2db8b" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/requests.csv", "bytes": 620, "sha256": "350d9912fd23a3af299c0080b541d24039034d6086532468d0d92fae23323713" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/requests.json", "bytes": 41392, "sha256": "69db7db6d863e99cb279cfaa8b193e66adb751a440e496a6f101f1582c62dff6" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/resources.jsonl", "bytes": 306548, "sha256": "5925c4363dfa4f3902899087e82b4ab69d2b4235748d3b4be481ae1ec90f6aea" }, { "path": "context176-47104-cpu45-f16-u8192-screen-20261005/summary.json", "bytes": 6334, "sha256": "642fe3113f99462113aa0d3bf4277c3ee2c9158173fe334fc5ba5cde967e7dc7" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "ffee4c421befbc2f9bb01336a5a00e386a02ef6cba5be48f523ab29641bab8bc" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/quality-44236-0/metrics-after.txt", "bytes": 2910, "sha256": "22efa693ac4661a06d34309b1a78ebef21c72cc6273625d247ee66cb404718b5" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/quality-44236-0/metrics-before.txt", "bytes": 2897, "sha256": "f97cc94c0e2f810c67cd452b4d30b8068f1e145563022784821bb7271bc4c487" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/quality-44236-0/output.txt", "bytes": 598, "sha256": "dde8c759308e7e50390c1d4f38f9969cba44d7d7b36f92ecedda517565ca57fe" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/quality-44236-0/payload.json", "bytes": 384822, "sha256": "9856632d0a5c42c04455ee92e81542f107d7d15a25522f9f96427c7cfbcc7b01" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/quality-44236-0/receipt.json", "bytes": 14214, "sha256": "fbed47bd1048533c4ab5ec375d0048dfa8c20022073be51330b0a234ac08473e" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/server.json", "bytes": 355, "sha256": "dbbd8ef6a37fd6f58f533bb2105e33c97aeda356c872d818e2d6ec8b96a4d290" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/server.log", "bytes": 52979, "sha256": "f8a5122501281799319913e5cc63ac4cc83af04bf755f96e5544f6460ad5ecb4" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-44236-0/metrics-after.txt", "bytes": 2909, "sha256": "cfe7b7e72d527d1ab1e775a4b6c30be924a77e33385f9e39cc25af80f89bd7de" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-44236-0/metrics-before.txt", "bytes": 2894, "sha256": "ce410ba491560f21e88d740ef02839d8078a0dda0ec6cff6cb52ca8afb510f26" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-44236-0/output.txt", "bytes": 1696, "sha256": "617b30227eb3e70f141f9d4af33a4e2cd55e705c58e84b0069eebe7fcf6b7b36" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-44236-0/payload.json", "bytes": 384833, "sha256": "ce0856a3f8a205de03c5739bd3dfe7494547e0f84fd6783054d288366d7c5ccd" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-44236-0/receipt.json", "bytes": 16293, "sha256": "cbbfeb26f4334273c6bfb4029960242383ac6335a61bfbd67fcf0bc6e7c627ec" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-8192-0/metrics-after.txt", "bytes": 2905, "sha256": "d88d4ed641be60a8780f1681a765a20b4df0c3899eafe2173237b4c1410a48d4" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-8192-0/output.txt", "bytes": 1428, "sha256": "7d7644d7dc9b35dc7849f1b144fa3aa061a5111f6ee1cc99c9555079d32d6e0e" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/ctx49152-cpu46-f16-b8192-u8192/speed-8192-0/receipt.json", "bytes": 8924, "sha256": "3b8d45abf72417182e1b211444bedd658b46dffc615e992ed24eff97d0f552be" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/requests.csv", "bytes": 619, "sha256": "6c1db6dcd6cb42e1f3aea1039dcc680e8a6f6734a9451bc1a715cf50965e64a6" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/requests.json", "bytes": 42225, "sha256": "bf6bf41fc09f376f8f2fdf312f21a55d62d1781c6ae80d2f904c31a6662413f5" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/resources.jsonl", "bytes": 303834, "sha256": "5e0e63069c9eaa484b2d43420d5e5aa11e56f8b8d6529f2ea6dbf37006981daa" }, { "path": "context176-49152-cpu46-f16-u8192-screen-20261005/summary.json", "bytes": 6335, "sha256": "d0765d2dc3e2a2b25b973139887d751a99fe0fd75b31efac1f04e4c3a7f5d414" }, { "path": "context176-65536-cpu43-q8-20261005/requests.csv", "bytes": 146, "sha256": "eb9e980aac7ace44eefe900bca780d2dc89afb76f9d4037baf7df4fd0ed1161f" }, { "path": "context176-65536-cpu43-q8-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, { "path": "context176-65536-cpu43-q8-20261005/resources.jsonl", "bytes": 1336, "sha256": "d0aed893e7403029e3fd7eb041277a2b2b68403ebd9f72c8a6dad3bfaf1a23e1" }, { "path": "context176-65536-cpu43-q8-20261005/summary.json", "bytes": 1262, "sha256": "b085574e228d2f1c21e3063c97b5c462982ae13117ac1174d7c00061399ecdce" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/configuration.json", "bytes": 2144, "sha256": "0c02056115b63d74a5cbb37a2f75254dda81f84e2d59424a8c8e002a6bed2def" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/quality-58982-0/metrics-after.txt", "bytes": 2910, "sha256": "6401c06946d4c646eec04835146454f0aa558a0a9d941664db3599431300003f" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/quality-58982-0/metrics-before.txt", "bytes": 2897, "sha256": "9c3e347e6d564d78dd52c3ca87c9ed69b22dcb917e80ac2bf2bdea5c70fb91a6" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/quality-58982-0/output.txt", "bytes": 993, "sha256": "c69ea1664534624eda11b0efed79fd3bfb92a50a91d3c6329406210d6831d59b" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/quality-58982-0/payload.json", "bytes": 512974, "sha256": "572c1564050798b282fc48051bef1e3ec30e11ac90afe889065ee2dcf158ff0c" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/quality-58982-0/receipt.json", "bytes": 18853, "sha256": "8e6933d6344be343295e686f4c859d80e1f824f892d52dfc37efd845b42318a2" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/server.json", "bytes": 355, "sha256": "675c6717bd6f5532e18e00bbe491b222107d92098b5badb888b00cb6700ca0b6" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/server.log", "bytes": 54202, "sha256": "ead7d5c5928e8b324ffde217a5c722840a5159344c7105b8d62832ae7ecd456f" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-58982-0/metrics-after.txt", "bytes": 2909, "sha256": "e9c2ef10fa49340b7c1c7f28766212924d25d2846dc7f1868d7ca201ac26bf3d" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-58982-0/metrics-before.txt", "bytes": 2893, "sha256": "c2392ff066d1226ab1f3ace7d2bc9a6d2fbbb4159ebd1be7a2f8397f20bb8901" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-58982-0/output.txt", "bytes": 1289, "sha256": "80c56c7166e7a3e47b1fefa869570bc19cf72afa921a3ea67ec14c37076b06d6" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-58982-0/payload.json", "bytes": 512991, "sha256": "78e328a59ac9d70e89d03579fe483bb77df64bdf43c60b3c70102faa18160ca7" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-58982-0/receipt.json", "bytes": 19775, "sha256": "774f92d706eb2ee12e36df9c520f1335b6e678c4ef0eaf635f7faa1992e6078c" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-8192-0/metrics-after.txt", "bytes": 2905, "sha256": "cfbfe67c67a6e22d8979fe50df4e1d4ff9a8f07b3ab8bed7cbfbd3644a3a6f2f" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-8192-0/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-8192-0/output.txt", "bytes": 1155, "sha256": "d4c5e15fb6c53179749bcad44a4373e7cdbb6bb4b1b4aff44721ba3c95641254" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/ctx65536-cpu48-f16-b8192-u8192/speed-8192-0/receipt.json", "bytes": 8527, "sha256": "a40c2e30cb80324f339c6e1908b433c5e6eeb99828bc7c582cafec186e1489a9" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/requests.csv", "bytes": 601, "sha256": "8410c086be35eba705b031413e3070cf34f323b80e3b3a3d6be9b5688af79c36" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/requests.json", "bytes": 50549, "sha256": "dc983c4ef8356d366d761ce45d9b5e990f887f0c425068b76374a43b05ff494c" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/resources.jsonl", "bytes": 375241, "sha256": "97be738f584169861d6c73003b0bbdad7259538249b9bc9d107014a32294a459" }, { "path": "context176-65536-cpu48-f16-u8192-screen-20261005/summary.json", "bytes": 6334, "sha256": "c0e45ce846e94547e8e8f0fff83e33ea68d34fd28b2c554122d3cddd12c3e5dc" }, { "path": "guided-speed175-20261005/baseline96/cache.json", "bytes": 359, "sha256": "334c94a97a774c6b7518dfbf2318989fdae207136f64b0bc65294db4f6faa752" }, { "path": "guided-speed175-20261005/baseline96/configuration.json", "bytes": 1622, "sha256": "f5f2a2fdad1dfa2182df479de95d4d672536266e6b3577edace51c083646f840" }, { "path": "guided-speed175-20261005/baseline96/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-20261005/baseline96/quality-8192/natural-quality/metrics-after.txt", "bytes": 2592, "sha256": "cbcaf8e10ef27f747ab79048ee417106a8ffd45d884646c068beb7d60976aedd" }, { "path": "guided-speed175-20261005/baseline96/quality-8192/natural-quality/metrics-before.txt", "bytes": 2579, "sha256": "b1b6a66734f919fcade85685b9d15cfee5a5348284db9099b04970df13293793" }, { "path": "guided-speed175-20261005/baseline96/quality-8192/natural-quality/output.txt", "bytes": 569, "sha256": "5f01f8b922e954c482409991ba4ea87ca62a52bdd6ccb8d1ac78ca3613c4378c" }, { "path": "guided-speed175-20261005/baseline96/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-20261005/baseline96/quality-8192/natural-quality/receipt.json", "bytes": 7277, "sha256": "0c0069492e2fd31f95f4dbd13b4a2a0290015e236edc4bd13270e2d387f368b9" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "da771a112bed8ac25258e58cfd37a475011c03c96b06f1690e1817f3c71d4fbe" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/fresh-prompt/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/fresh-prompt/receipt.json", "bytes": 8951, "sha256": "475520653eda0a22efb2a9f80c355ad854f5b0804856290e1667470b01083940" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2590, "sha256": "8b6c731622728c8edd12f6eb7ebb5f92d0fca895fe64bbd5d4ec4711adbb2e8f" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "395d7710039a9c9e3c4db035ba74e0c5d0a0c0bf4b8c949cff7f231207c77166" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/reuse-requested/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-20261005/baseline96/repeat-0/reuse-requested/receipt.json", "bytes": 6246, "sha256": "62ee983d8110ffed9eb8943a9ad1278e3851a516a73ae7d6b4a86318e37291a0" }, { "path": "guided-speed175-20261005/baseline96/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-20261005/baseline96/server.log", "bytes": 35386, "sha256": "71d23d31cd1f23f8f485aed62f256d9300208bb63ccb8f026b97a731c5357e2e" }, { "path": "guided-speed175-20261005/baseline96/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-20261005/batch8192-cache96/cache.json", "bytes": 359, "sha256": "334c94a97a774c6b7518dfbf2318989fdae207136f64b0bc65294db4f6faa752" }, { "path": "guided-speed175-20261005/batch8192-cache96/configuration.json", "bytes": 1622, "sha256": "087814832b2ab620091c561a4a98892c9bb19f07e6ad8c34d8a0167e031f4cd4" }, { "path": "guided-speed175-20261005/batch8192-cache96/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-20261005/batch8192-cache96/quality-8192/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "59e62d8e940165ef3d8ddfb02ddd8c0fb106343d16a072da6b940289c74851a6" }, { "path": "guided-speed175-20261005/batch8192-cache96/quality-8192/natural-quality/metrics-before.txt", "bytes": 2580, "sha256": "5a14db4adc59d78c7e269c0d421e44a999639f91616eae147ae193002bf3ed3c" }, { "path": "guided-speed175-20261005/batch8192-cache96/quality-8192/natural-quality/output.txt", "bytes": 484, "sha256": "5e615214bd41ddf9e63435e710f0e8b98aae254f7ac38261236ce7f268c5a252" }, { "path": "guided-speed175-20261005/batch8192-cache96/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-20261005/batch8192-cache96/quality-8192/natural-quality/receipt.json", "bytes": 6516, "sha256": "ea59405c687ef8d8fbfc76413946ee24f8ee098a79be9d0bf7a3bf35eb951213" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "250ee4e43df1dbd7d9fc7918c95e1097c36c68995cd29dd0e709f3556618c3f2" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/fresh-prompt/output.txt", "bytes": 1120, "sha256": "53ae6cf14377470e9ab4fc665874f5ba803a0e796d52c395751e43dca2e68f54" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/fresh-prompt/receipt.json", "bytes": 8583, "sha256": "53c65e7b46f5ca89cffd8bd09171f05efd30b1e0c38cfea72176a85e56dcc456" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "30642b69bf9ae4f74a8ac054c4230b29f96c73b921163688f2257cef2dd5c631" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "62dbe0a279e02857d08fe802017c8ccbb43b601d2c25a25a10e52528b32fd53f" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/reuse-requested/output.txt", "bytes": 1120, "sha256": "53ae6cf14377470e9ab4fc665874f5ba803a0e796d52c395751e43dca2e68f54" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-20261005/batch8192-cache96/repeat-0/reuse-requested/receipt.json", "bytes": 6648, "sha256": "81ff2503e7fbd5a4a6291001fcfa13d37cc28199f69280f2d90687b70991a87a" }, { "path": "guided-speed175-20261005/batch8192-cache96/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-20261005/batch8192-cache96/server.log", "bytes": 34393, "sha256": "44fd8cb5841098041aa203227ceb05df95f34c29fd39e0165eae90882b8374fa" }, { "path": "guided-speed175-20261005/batch8192-cache96/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/cache.json", "bytes": 458, "sha256": "9f7a7f059d845d059141c0671c536f58a114552f9cbef96b560ddd2bb2941c67" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/configuration.json", "bytes": 2118, "sha256": "a485cd73d5fda457a43f623ec7551a264d41533b6c75a30ea5094713144b20d4" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/quality-8192/natural-quality/metrics-after.txt", "bytes": 2911, "sha256": "39effc8b1e28e1a042af3e87742d17d20aabf811a3ebb978b9ca8301fa958be5" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/quality-8192/natural-quality/metrics-before.txt", "bytes": 2898, "sha256": "b52efcd5833a774bf4cfefb0a20bd534a38bf495a604cd0ae5c45db54b9778db" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/quality-8192/natural-quality/output.txt", "bytes": 833, "sha256": "d7ccff64be97eb58a0cc266640bbb6a72f9da940e95735efa9f500e017419440" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/quality-8192/natural-quality/receipt.json", "bytes": 8232, "sha256": "0a672aac538133e282ac64fa605bd56a51dfebcce1e1713b3b000e023106b226" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2906, "sha256": "c3896dc1b61b307ebe91b89069f816353971318686998666ee50e52965a1bed2" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/fresh-prompt/output.txt", "bytes": 1324, "sha256": "9bf1ace2e161721f5545a2b9bfd221bd76a693a8b1967cfa0fbda60d52bd4411" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/fresh-prompt/receipt.json", "bytes": 9527, "sha256": "a0dd7c28c11331e9d8877420f805606a7975fceb7c877b01715bc0ebfc224b64" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2909, "sha256": "599b932b83d289ee1f3f08f082d80faf67af6bddc18522d49d766bc0b4708d77" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2894, "sha256": "f4cd86edf600cf04b19ba87df436db6d0c9c4aab905debe7490c941559d5fdfd" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/reuse-requested/output.txt", "bytes": 1324, "sha256": "9bf1ace2e161721f5545a2b9bfd221bd76a693a8b1967cfa0fbda60d52bd4411" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/repeat-0/reuse-requested/receipt.json", "bytes": 6431, "sha256": "9018fbffe02f0f62d5505da114ba6b9e30d124215eeca4685a23619519ff13d6" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/server.log", "bytes": 51235, "sha256": "be48c0be869177148c4e442bfbca0a7ef43862df5d44ea8388256145641b5eed" }, { "path": "guided-speed175-20261005/gpu-mtp2-cache96/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-20261005/overlap-cache112/cache.json", "bytes": 362, "sha256": "acbcd84a1cc3907c1ce465c7ada50b43a0dcedc94ff058b84a3d8730ec936ed0" }, { "path": "guided-speed175-20261005/overlap-cache112/configuration.json", "bytes": 1595, "sha256": "741ea1c27bf242763b06503282eaf126c7d27035a18d3c65e991f6145ab75b84" }, { "path": "guided-speed175-20261005/overlap-cache112/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-20261005/overlap-cache112/quality-8192/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "9cabe1fdd23f936e8038c23594e8af71719ac9ffecd5f11f0d23be426d2a0fbe" }, { "path": "guided-speed175-20261005/overlap-cache112/quality-8192/natural-quality/metrics-before.txt", "bytes": 2580, "sha256": "237c3f2dd07a6862cee26d901a408d60b0b50694adb6068dbfc9da99c7fcb826" }, { "path": "guided-speed175-20261005/overlap-cache112/quality-8192/natural-quality/output.txt", "bytes": 584, "sha256": "194d0040a6d8828d1ccafa15e176ddbce7b3d3088232510ad1d231ba9aeb06b9" }, { "path": "guided-speed175-20261005/overlap-cache112/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-20261005/overlap-cache112/quality-8192/natural-quality/receipt.json", "bytes": 7673, "sha256": "45b5d02f90d338718e1560036486abaa6c318703d1d3606d5d7b609d25f085ed" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2586, "sha256": "a4a77774a3914f8107b05ab47fa2196fa4d7720fbbebd992dca8cd66e65a3a8a" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/fresh-prompt/output.txt", "bytes": 1331, "sha256": "6622ede60f66d243dfa53f17d0a492e13a63f90b1bbe0a5d727fd5c765fe8c64" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/fresh-prompt/receipt.json", "bytes": 9734, "sha256": "ebde97c98bf6a0d0a5a40eddeafe100f4ebd9a8a66b9bdc52df44cb55cb154fb" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "a6548b3ce71ba315f4ee0e4f288a9f8e5c0d35484b347827d1512d2e1c7e17fa" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "06a5446f0522f03768d88e4b07e7a31b3aade44b51fcdfd78fb7b75124c843d2" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/reuse-requested/output.txt", "bytes": 1331, "sha256": "6622ede60f66d243dfa53f17d0a492e13a63f90b1bbe0a5d727fd5c765fe8c64" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-20261005/overlap-cache112/repeat-0/reuse-requested/receipt.json", "bytes": 7034, "sha256": "07cc892ffca14e23a755d590a95d46e97e492a7cea77a5dd9ef7158bfa32bd70" }, { "path": "guided-speed175-20261005/overlap-cache112/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-20261005/overlap-cache112/server.log", "bytes": 35703, "sha256": "6b05c083d99501c33ba687073ab62a4bf517ce66f315ac249ab3afc22db87f52" }, { "path": "guided-speed175-20261005/overlap-cache112/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-20261005/requests.csv", "bytes": 1591, "sha256": "d79151d4871dafc560c172fbc02d78374a1a5b99ad52953aa16ee38099c18d27" }, { "path": "guided-speed175-20261005/requests.json", "bytes": 99267, "sha256": "d8ae6ffc5a8dbd34ba7ab6c0f20b1c4d38c8194e7104473eccc2a37b5c20752a" }, { "path": "guided-speed175-20261005/resources.jsonl", "bytes": 434539, "sha256": "a4d54a790ff274e00c274febf119a800cb369d78101501d9fcfb183c47dd9494" }, { "path": "guided-speed175-20261005/summary.json", "bytes": 14075, "sha256": "ea42c156b8ccbf351b9b057c8542a4390366b7cb982b9cc4d98339f32625afff" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/cache.json", "bytes": 478, "sha256": "c0ca804e71544276c3284af1c0d4219055b8ae332905fb716447beaa9dafff86" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/configuration.json", "bytes": 1650, "sha256": "fb9c4186871f22d7bce37aad37f53dcfa9fbf0c4bda8f66d124bee4c38e959db" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-16384/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "683ba1e9ae06be10e4de5743d25950abca96d7849068b66825082b650390f2eb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-16384/natural-quality/metrics-before.txt", "bytes": 2580, "sha256": "e5108af7906cde2e61b4c1bf2c7be2af370f1e20b5da1ca394301be635b1879b" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-16384/natural-quality/output.txt", "bytes": 639, "sha256": "632148770e173903ae34adc1439be6d1bccc4369b2d724d623c68680cfbac1b7" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-16384/natural-quality/receipt.json", "bytes": 9228, "sha256": "2909d4f6df35874f40a39ac4f170ade148e8a2123efe0ae32f83cb9141b49d6c" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-28672/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "1f5ccc50083235d0cc8cde2b141b2fc424c8396dfd66142b967b10ebf3ccd282" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-28672/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "99b5ada7094c5c4a29ef882895df44b3a1a9bca6320aa083228916e14a90a848" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-28672/natural-quality/output.txt", "bytes": 803, "sha256": "5a0a9b2a0095da77ce12223f37c3f5bab27755bfaa6393d6a8c8ed44a71465f7" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-28672/natural-quality/receipt.json", "bytes": 13478, "sha256": "a631cc54f5c2302f60492b1f854634a94e3dfc1eebbc571b2513068bf0323893" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-8192/natural-quality/metrics-after.txt", "bytes": 2590, "sha256": "cca9f1d3eeaa722eb160fff06b014dc35ce5c6a5fd87acce7ff94ad888f9cc32" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "a0d87239e63691c88ff89098b047a3c1ea165e0656e675c2025ac75fa39e62a6" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-8192/natural-quality/output.txt", "bytes": 721, "sha256": "025800792ce2322d35b6301fea4ecbf635cc1ee1307bfc51f6ef9d2cceb4878b" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/quality-8192/natural-quality/receipt.json", "bytes": 7296, "sha256": "8de8c74aba61e7ca688333e8481d478a7ad2ac1a9bbfee1e986453c942daf7f4" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2586, "sha256": "82bd8c76353d976b0f6d978ad4ca1c93153840d0fd3e1644843bfe8a5b81476f" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/fresh-prompt/receipt.json", "bytes": 8584, "sha256": "900c0e4aa850654d2a02d372dae792f7c03472791bea20a8c11b2ecaa45562af" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2591, "sha256": "4a5b8206fe28cf6ae2aa2f9681ed1511d3614294d3720e34995072f7bdc6c0ae" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "60afd6a7076e2fbe84ea8a0ab58d816f9e7257a4764aba0399a550df4c1cefcd" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-0/reuse-requested/receipt.json", "bytes": 6653, "sha256": "680d34f872f185d845f0220a893248a7f397bbe0c3790cd244ce945b10225d1c" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "82059fc4ff2ea029adcd7585e075ccf6f4bf8edf0e2681de37601e82c718acfb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "96317eb101e6a2ec12f4d6e077e40cd5f99149314b6ca6e2d2062c25ca601ab9" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/fresh-prompt/receipt.json", "bytes": 8591, "sha256": "44bbcab53006938412672ad93b9148651cd8f286654ae9eb670194e2575329cc" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "25000aa5e77727d3c571c10de5db6658c551b48d077de4f43f2e42b48f1d38f6" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "830583d7e7530008a53bef7eb23f2ae496a8bc275ce89c7f4e5a7b62dd1a2a55" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-1/reuse-requested/receipt.json", "bytes": 6658, "sha256": "35f97460d861129722a266a228f5fe1bcb033089210049aef33c2c0e707a7609" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "d8817af12a0a3ce7a71ca76e99c83915e24eb0acedf48c0443aa174946f32d4b" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "166f7aec39a6c59732b13e4ebddefb45294cbeea51f18686e3f2cec869b05e49" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/fresh-prompt/receipt.json", "bytes": 8590, "sha256": "0690e4a8c6f69429b22fcd8d4142b4a516646809ad7525ecf80abe08a2c4f020" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "8c652082bc62011c4c55ed4783a9db66a1b9cba599f2d81fe44f2094b2173369" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "ea5183b890189c8567142aee2147ec92dba9b3ad869dce6b0a69a83a2cb1fc8c" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-2/reuse-requested/receipt.json", "bytes": 6659, "sha256": "5331e0bb08d0391ccadf33a8786d7d9e6f19eeb1bd542309c0fc35374707ebb7" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "6456b54c86bd2f3ba8372018dcd9d27570f459e8966a76d951b7db37e1a8c0e1" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "82b65cdd833dc3c6a5a99bf7d11952c78542d55aa6eae63e2f5212b8399ae266" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/fresh-prompt/receipt.json", "bytes": 8591, "sha256": "202cc8dffe97249720949b1277845cf2197fa98058cd339fa6e5bd4e4d0368fb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "496970df481d9c28dedbed09d525326f6c699eb946e8bc69dbf21e3f3fdffebf" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "36a2891f518095a35b1d4de4e18a83fa4b570e12d2bac23e40dad24ad277e038" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-3/reuse-requested/receipt.json", "bytes": 6659, "sha256": "c3fffd2eb2adc2a5ec557b06273c902ffda5c843046098d072ecf6ba6bd4e8b9" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "36750ef93563970b3720687bcb1a2e26b1be90480df8344ce6878b7530e80020" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "0f90e1b2d7ca10f02cf8000807d9b2720067912f96f46cd0cf400856cf35f280" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/fresh-prompt/receipt.json", "bytes": 8592, "sha256": "baec3d19b433102073c21d1b847ca586f34fff4006e28dd76f445f01bc585436" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "16e8a8a2a8fe77940a8337e20bb6562c862c8d1c27d46c42bab6f3e3032d4402" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "d342c0e2946a0ba145e8e5a9da14a80336cf0c9e8e4f0a8fefeace7fd8e989f2" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/repeat-4/reuse-requested/receipt.json", "bytes": 6656, "sha256": "2ebd8f3ac6f148f27acf971a01d1287491e4cf7c32b2aa67133a85d3c16be12a" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/server.log", "bytes": 72067, "sha256": "486930a5f895f89391cc0f91c614c94793017e21cb1bfb82b47e7ca8246815ec" }, { "path": "guided-speed175-finalist-20261005/batch8192-cache80/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-finalist-20261005/requests.csv", "bytes": 1655, "sha256": "8e9a3d97583ab60ace8d88a279907bec2649c63deed46ea970d2f91948918674" }, { "path": "guided-speed175-finalist-20261005/requests.json", "bytes": 112139, "sha256": "574d3015593566462efe2d6e0554b087e3cee0208ab1f24715bf53bddee16bad" }, { "path": "guided-speed175-finalist-20261005/resources.jsonl", "bytes": 478271, "sha256": "086cb1475606d018ccadd1bdb12c4573cfa0b55d07d4a792ca0cfda2e2da27c6" }, { "path": "guided-speed175-finalist-20261005/summary.json", "bytes": 7727, "sha256": "285e6cd2c9ddc912cfef0d28e59b93c2bda9a05419ee22bff25e6b5407404a33" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/cache.json", "bytes": 359, "sha256": "9d93352decab35d2fe5b1b843056f8821bff064d797372714847dc6d12ad0eec" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/configuration.json", "bytes": 1622, "sha256": "6b8f16237300e7125def1b702d0931a9956af3fde08bcb92f027eb4c27765a71" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-16384/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "d00d7e1cd7fcf86d01b89a661635ce6769fa8dc1b60e46a9cc05340050e6d5a7" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-16384/natural-quality/metrics-before.txt", "bytes": 2581, "sha256": "4458b94b540d63219b15b1f161910cb361f89e2cc679a3b2fca03622c35a488a" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-16384/natural-quality/output.txt", "bytes": 547, "sha256": "3aefd4613c25c6b2ffe1c430ba31428a74ebfd557ba5002b67b9759dc0a3b899" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-16384/natural-quality/receipt.json", "bytes": 9230, "sha256": "497937d700a7a11c653bf45f551e66987a2c037f4bdb8ed96729d2d3dcf1d704" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-28672/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "ffbc7db4e6b8779899aa9febc827fe722e14de7b06688fb69bf2ba0ced75dfe6" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "6b3e1199e4b828502f56eacd7fe70c571a4ea5b807ae2629f9cab4d5c597dfb0" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-28672/natural-quality/output.txt", "bytes": 602, "sha256": "10b632bf2a589367018c16e2daf5412af8ef35755f4c8a3f41ac185d345b9f2e" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-28672/natural-quality/receipt.json", "bytes": 13095, "sha256": "63d81c3098e9ebfe4bf9f322d4d127a6b85ea72c8aa6e4516c8cfdd9f956a577" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-8192/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "00b444c87c81d9dc740148c985cfbcd1f0cc92cecc10f4d94e257f7790dff5dc" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "7096a7b4035f4df402fc447e0facd8e312f7fba9284657c956b47b9c7f4e8cda" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-8192/natural-quality/output.txt", "bytes": 674, "sha256": "c4e63242e43455997c6da06c6d1f363f948592651947cb1f5c9ef9d1fbe0baca" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/quality-8192/natural-quality/receipt.json", "bytes": 7297, "sha256": "dd91be30b6587ee6893a91910f4e4b85e1c9d5156d58307f6ee57f7b0752dfb0" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2586, "sha256": "837238abc871ab3c7bcb8bdbcca495014273f5ae8e9474cfb9d652a6c7b18f85" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/fresh-prompt/output.txt", "bytes": 1265, "sha256": "f9d01c5df22465d0fc5d839da1abb7a10dd2c343e88773243205b081ac38db29" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/fresh-prompt/receipt.json", "bytes": 8973, "sha256": "bb5d7e156bb261dd24e8e5081cc9ae3f5464b98a8961df3f80619ccc2877c12d" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2591, "sha256": "b021c65c3a427cd0ee449fba6eef0b398125409cea99e99089477a898a2a7274" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "dd0134face2fb523569314cd034568676f6e7471e821b4d8b1d126204a8764b5" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/reuse-requested/output.txt", "bytes": 1265, "sha256": "f9d01c5df22465d0fc5d839da1abb7a10dd2c343e88773243205b081ac38db29" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-0/reuse-requested/receipt.json", "bytes": 6656, "sha256": "92ebde32be90ff2ded6aaaca944e3072279bfb67dd94d1690f306d1d3680c4cd" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "0c60e935c72ea3a31bd5dc973a536748f5028a255c2ae5cdfe82e95e5412d7f0" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2579, "sha256": "5ab0c95425cba9b3098a62df18a44710204e4b60374ac82c05d8b3c748f7b6e7" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/fresh-prompt/output.txt", "bytes": 1265, "sha256": "f9d01c5df22465d0fc5d839da1abb7a10dd2c343e88773243205b081ac38db29" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/fresh-prompt/receipt.json", "bytes": 8589, "sha256": "f7fefcd410de185af773b0c073b7ef7ef55de780edaff66e6ac70690ba224b85" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "5c6c891873cbf8491796613ab3822b75ef47289dcf6512f3e85c8c99955bf75e" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "a385c21d27f5aa7bdcdbf708279de63c826af180657ff05595180b4701352217" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/reuse-requested/output.txt", "bytes": 1265, "sha256": "f9d01c5df22465d0fc5d839da1abb7a10dd2c343e88773243205b081ac38db29" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-1/reuse-requested/receipt.json", "bytes": 6654, "sha256": "3cf27ebd00acb109522c33ddea01af200495ce06c55551d20aed90e0424403ff" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "bd43c4d9fe9d3cb8fefa108e2169552d655a633a5dd81e806d3bde2daad8ec9f" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "5ef756a088700d67ee68fc9ef3638eebd0136bcac7b9a98edccd316742e88f7e" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/fresh-prompt/output.txt", "bytes": 1265, "sha256": "f9d01c5df22465d0fc5d839da1abb7a10dd2c343e88773243205b081ac38db29" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/fresh-prompt/receipt.json", "bytes": 8589, "sha256": "b17f4592da17360191a21be7194908664af0d79e12e2e8eec40536c193b17595" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2593, "sha256": "cecd1c700a0fb86798e1133fedb26ad56b27d3ab7a3c4e457da254db87d766ae" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "6d6187b600d6024515db9da2f7d2a82fb34becd4f8e2d1d11c0c4bdb7117d27c" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/reuse-requested/output.txt", "bytes": 1265, "sha256": "f9d01c5df22465d0fc5d839da1abb7a10dd2c343e88773243205b081ac38db29" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/repeat-2/reuse-requested/receipt.json", "bytes": 6652, "sha256": "c7a0cbbd73b343dc8436d2be4b05e698be886d4cba8b5b16d871c59d6e94c70a" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/server.log", "bytes": 57921, "sha256": "858b8da237e589077bb615efbe84f2b023fb0e600010b08f3da9e36d6d41929c" }, { "path": "guided-speed175-followup-20261005/batch8192-cache80/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/cache.json", "bytes": 458, "sha256": "56625906c4a2321c95649fa45ae3df710b8656357017564da9fed894ddfa9a7e" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/configuration.json", "bytes": 2118, "sha256": "f737871efe0e520a93facd3131f74a9e684b9f0a970904afb9ec7b8bf343a661" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-16384/natural-quality/metrics-after.txt", "bytes": 2918, "sha256": "764e46df1475b4b56748d923d9d03c3fe7e036455b513ba30d9679dacdda1f30" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-16384/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "438fc2bb4cabbb9717d71730cd3440b20a432a70c6f5556d37f298f42fe2bf0c" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-16384/natural-quality/output.txt", "bytes": 575, "sha256": "a9b96f998de6e2963d3b3e8552d3e4081d6864e6f5af4cbca2755e81fc236744" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-16384/natural-quality/receipt.json", "bytes": 10929, "sha256": "6e560ffdd01fc4ca340fb4368f9959b93460bff6765f2ec0a7b81d5726f3a4ae" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-28672/natural-quality/metrics-after.txt", "bytes": 2917, "sha256": "faac3952c4c9dce602ce9852c5c3c47237bfc011b11617404b32b1fb45a247ed" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-28672/natural-quality/metrics-before.txt", "bytes": 2906, "sha256": "80263effa34ca4271bfdeb0dba48b0ef4c5f54bcc18b09c35c6ac0bd34722d1a" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-28672/natural-quality/output.txt", "bytes": 841, "sha256": "3b864d9dab6a61975b8d82ff3c8b87049358ec74837d2e8116e1a1ed6a56bf4b" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-28672/natural-quality/receipt.json", "bytes": 15961, "sha256": "7105131a40f9c522fb68eaddd39acaefce420289c2632c9ce175a0a4bb61044d" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-8192/natural-quality/metrics-after.txt", "bytes": 2917, "sha256": "a53f03c16dafbcb12b8606dc66684dd145c65061419bde24b37e6fc467046f08" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-8192/natural-quality/metrics-before.txt", "bytes": 2904, "sha256": "c13f2b180dc70012e3444f577911b61bb762cf1b9c86beace8f29c4e3c262d40" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-8192/natural-quality/output.txt", "bytes": 555, "sha256": "1c95fc87c9ba1762b0fd278874339e41e28fb4becbbd09953fc919683b125e80" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/quality-8192/natural-quality/receipt.json", "bytes": 7850, "sha256": "c784012ab7d97d2c6c543e2d92b00b0ad8bb725d373897656dc728103d9d6426" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2906, "sha256": "6a6bdfe7022dea628f5cccb386f64d6556cee4418be2c4b58851095334d0b97c" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/fresh-prompt/output.txt", "bytes": 1109, "sha256": "49bdd08ac07ec7303ddf1a1cebd64af9671812f94d1100d31e147926dfa55759" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/fresh-prompt/receipt.json", "bytes": 9533, "sha256": "a94206a9cca69eca5b50e337eab2e5f42ed1d8cc48ff45a56b996dc21f6e8435" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2910, "sha256": "d8e0fa4ccdef0fb4694e13ea6b72bf7af9d3ed8af963fecd16b6bec857699728" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2894, "sha256": "98ee41aa87f6c875a4c5ebbe87a11289c47e2b55e7718ccca15e0d753c40b825" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/reuse-requested/output.txt", "bytes": 1109, "sha256": "49bdd08ac07ec7303ddf1a1cebd64af9671812f94d1100d31e147926dfa55759" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-0/reuse-requested/receipt.json", "bytes": 6824, "sha256": "78f1c96fdcf42b1ba4c2c33f4c693dec462d2d81a068495526090cc595f62edc" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2912, "sha256": "0f72c936bcc7fbe9462c80103d2f27bcabe269ba27881a63026990befdea36e0" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2898, "sha256": "54e27608846b1190a28a56892e376c9741c0d743852e9b9b511b04c2dfd2ba8e" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/fresh-prompt/output.txt", "bytes": 1109, "sha256": "49bdd08ac07ec7303ddf1a1cebd64af9671812f94d1100d31e147926dfa55759" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/fresh-prompt/receipt.json", "bytes": 9531, "sha256": "4b639d9ed1739dcbbd3246b501fe157ef439fef485705098a0e734699ed82c3f" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2914, "sha256": "97dbfd5de6da6674d4a63283a40dda9ce6cc3835a92878a6eeae43b4d33927c9" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2900, "sha256": "23ef65b539f5bb139c8c05229394bee8c9c28961d954e5d5caf25b2d0267db8a" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/reuse-requested/output.txt", "bytes": 1109, "sha256": "49bdd08ac07ec7303ddf1a1cebd64af9671812f94d1100d31e147926dfa55759" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-1/reuse-requested/receipt.json", "bytes": 6830, "sha256": "2f8041c7f54262db1abcb9587e169b8aaf6f3d81dc52a7427244134e82cc2ed6" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2916, "sha256": "7cab481edb4e7e3fcc5e1e4acda6b8af1d8aedcb746adbec74d65af7661b78c6" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2902, "sha256": "48f936ad34c2eae8c1f3cfb6874246e6d926e152ac3e7a1a00badcddb722e6c9" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/fresh-prompt/output.txt", "bytes": 1109, "sha256": "49bdd08ac07ec7303ddf1a1cebd64af9671812f94d1100d31e147926dfa55759" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/fresh-prompt/receipt.json", "bytes": 9536, "sha256": "7ce3d41b4d052a7d4b894676c7dc2577c7ecc82430c948ad1f8ce21fddabe061" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2916, "sha256": "1df1ca705887a1b1653844f9ff2b0cd4068c7c0d7d600ab50d1819b837bd836e" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2904, "sha256": "668f32b51cfe9e2c8503f1fb2db94a3b26a7ed3d9dd227e36b46a6abc8da028a" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/reuse-requested/output.txt", "bytes": 1109, "sha256": "49bdd08ac07ec7303ddf1a1cebd64af9671812f94d1100d31e147926dfa55759" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/repeat-2/reuse-requested/receipt.json", "bytes": 6831, "sha256": "d4b1e09d0daae13bd3bb691e73dbe396a744e91cccc22aaea2be827cc7ebb2b3" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/server.log", "bytes": 79839, "sha256": "ae93511a74f05120bb286d3b20899d20c42e1a7553ec6f80257287b01a3758fd" }, { "path": "guided-speed175-followup-20261005/gpu-mtp2-cache64/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/cache.json", "bytes": 458, "sha256": "dc5a1fa225d3c786b747d661c9190fd7a5d84838b809b3901c45123abd9513e6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/configuration.json", "bytes": 2117, "sha256": "59e4fe17c871bf55f6b827a75c62acbb3e7b68e5c06bc39d8c82d88822679adc" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-16384/natural-quality/metrics-after.txt", "bytes": 2916, "sha256": "fb7a708006f684d7644957bec6461b660dd6993e182af53da5f5d2db78ad837c" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-16384/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "4f1a14e6c124ff290207f9e3b5070d1f9cfe0d1a0a5b08ce07f580b2af52b5f7" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-16384/natural-quality/output.txt", "bytes": 554, "sha256": "f7d9451b28eb5fe679a07a6d80d2b04c16a05e134dfdbd0127a780fc6c43d337" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-16384/natural-quality/receipt.json", "bytes": 10185, "sha256": "bbf93a67d279b8a93a24f6d975692bb943738e3bd922ce3af5217077ef401760" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-28672/natural-quality/metrics-after.txt", "bytes": 2919, "sha256": "c9c7c7fa5f4397e52b2305f7904e1bcd5e2f24c38e608e127f38baa18067d56c" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-28672/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "f00ae4b02bf6baac54f1bee3759671958c23a0c6b957275b7d86f3759eb77bb5" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-28672/natural-quality/output.txt", "bytes": 729, "sha256": "e9284150c43b84d4510caad263e2cb8962eabd57d606e3b1ff669b178909d39a" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-28672/natural-quality/receipt.json", "bytes": 15207, "sha256": "4dc11cc9853715062557ae9657ea177bc23da44f693aa0fd8b361d0835003220" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-8192/natural-quality/metrics-after.txt", "bytes": 2916, "sha256": "26388aca18bcba8f8795a74320896378f5c2b04c5a7c246ff2360efcec9af485" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-8192/natural-quality/metrics-before.txt", "bytes": 2904, "sha256": "0af188167570e87c01d5fca7d2d679016076faeb44519450bd41138e99341ff8" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-8192/natural-quality/output.txt", "bytes": 729, "sha256": "9c0dba6db3c763e64d6fae22f3294b80cdf2fda58c5af36b7be24d5d7bce16d9" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/quality-8192/natural-quality/receipt.json", "bytes": 7854, "sha256": "0a2a3d64e8bd8f2f72192d6f7a08678452691d39a8a83a1ea3c64100001a83f8" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2906, "sha256": "bc19e89411e32195b891f7047baf3e1ba07494284ab41f4e41dccf01e183f68c" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/fresh-prompt/output.txt", "bytes": 1205, "sha256": "bccd695d0aaa836bb3c1541cb8b036fd5f42b364cfabfba9bec7458cc5a568f6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/fresh-prompt/receipt.json", "bytes": 9938, "sha256": "f0007e52ea2107bd215ab406e1597dc871d5062e902c9458aaac036a117c5322" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2910, "sha256": "c9e3b7439c2359caff908e9370b83201f8b98551f80e0e57bef5d361bdf10bdf" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2894, "sha256": "6c6d2977cd399fe3acb5fc15fb07c4f624d530b668dd39d8735d3b780c860180" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/reuse-requested/output.txt", "bytes": 1205, "sha256": "bccd695d0aaa836bb3c1541cb8b036fd5f42b364cfabfba9bec7458cc5a568f6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-0/reuse-requested/receipt.json", "bytes": 7229, "sha256": "8b32458a0a624f553ac7d4f137bbf7e4254f0f6a74a71f3786762dbbf7e36c91" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2912, "sha256": "61932bb85841536cb7b792e642c456fb93020f9b37ef9694905ed9023f2ea269" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2898, "sha256": "734ec61b004d9cefaa4c8c732377ca35199df38375216a871ddcb50d12a5cc3b" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/fresh-prompt/output.txt", "bytes": 1205, "sha256": "bccd695d0aaa836bb3c1541cb8b036fd5f42b364cfabfba9bec7458cc5a568f6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/fresh-prompt/receipt.json", "bytes": 9546, "sha256": "e05052921240662553aafce6b20726928d568fe531e1a57c09f4e02757f79e0b" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2914, "sha256": "4cb28e3bd5a278730072ea16c6ed51acefc7fb20503579585636f747adb329f2" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2900, "sha256": "5082e5223290241b181d2f3a9609f0d8ffa6427b80e9ba78072323f8d9013aaf" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/reuse-requested/output.txt", "bytes": 1205, "sha256": "bccd695d0aaa836bb3c1541cb8b036fd5f42b364cfabfba9bec7458cc5a568f6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-1/reuse-requested/receipt.json", "bytes": 7228, "sha256": "6f54f0d6ea075a31a94565ad61842286b89c3720f3a73d707714fd8b5dae681a" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2916, "sha256": "78a090f830de083c826852ed948112914ca41ed8653893246671f2655d717f4e" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2902, "sha256": "28ea822e064843ac38c0a9764ca549284964f58e77b756227bb506f1c5e83997" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/fresh-prompt/output.txt", "bytes": 1205, "sha256": "bccd695d0aaa836bb3c1541cb8b036fd5f42b364cfabfba9bec7458cc5a568f6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/fresh-prompt/receipt.json", "bytes": 9551, "sha256": "d4ca7b9e413f945647d6a393fb997226d53c8e76f310ae90100e03ac837fb69b" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2916, "sha256": "cbd9ab61b3ba1b634c1acc6fb21f3fc57da49f4c34bbc733719efd9263840eb3" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2904, "sha256": "c2aa3f547744b283d95abc1c331b820874e4f7604d4eabe93ae27599af2fea65" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/reuse-requested/output.txt", "bytes": 1205, "sha256": "bccd695d0aaa836bb3c1541cb8b036fd5f42b364cfabfba9bec7458cc5a568f6" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/repeat-2/reuse-requested/receipt.json", "bytes": 7227, "sha256": "79490d1ca14f39fec5000f1c8cd06a068fa826ec1bcba933ab32a2cdb3b30431" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/server.log", "bytes": 77157, "sha256": "fd737654451555d36bace5d1cc9bc162422a429caaa8b56455b19b67492a3cd4" }, { "path": "guided-speed175-followup-20261005/hybrid-mtp2-b8192-cache72/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "guided-speed175-followup-20261005/requests.csv", "bytes": 3762, "sha256": "71976296c7eb96dac82c543ce988522484df4f9ed2000381f8b09e312298d201" }, { "path": "guided-speed175-followup-20261005/requests.json", "bytes": 260325, "sha256": "2d4ed717d23947939b9b51e7f271a4df3d477639e79e895a09c1ffd882d053cd" }, { "path": "guided-speed175-followup-20261005/resources.jsonl", "bytes": 1023813, "sha256": "693aa3c8b200d890d50ab9fad11282e06e6dbb2f1dba3858a25227eaf42200de" }, { "path": "guided-speed175-followup-20261005/summary.json", "bytes": 12706, "sha256": "aa8a052a4d5da522dfc392a06c85a8b7e242ba2b9e74ec6766f0b0e6dbe0ed13" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/cache.json", "bytes": 478, "sha256": "6408ece3caf9b78593564eee8af50fc9fa48a73b8c106b498c296b53e592f478" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/configuration.json", "bytes": 1652, "sha256": "1db4a961601d87d10a75d3b22ff824e977d6c7ba9224914d67bb18e2f174fb5d" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-16384/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "0500f1a54326d357c47b8919f0eee9b766f3b24be49c09a2ff6b55d6b2ccd723" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-16384/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "ac04dd56c3f08f71cb7df0d718014e091cdeadc3dffda47c5806ea0064049f66" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-16384/natural-quality/output.txt", "bytes": 561, "sha256": "e111b727b8701d046cdc8e8756c39d485cc782490c95d0aec24b3f44089c0dc2" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-16384/natural-quality/receipt.json", "bytes": 8882, "sha256": "34c42e2d07d54339ad297390d4e32a0d4d2238e2a1194479e922e8fa4e657d3a" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-28672/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "2ca667fb27a0a53fcf8da0a752ecb9e214a383fb669b1856b4e76ffd1d4923ed" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "52aefb78ed0e90cef051951341e0215c8e74fd6f097dc3c50bb433d08739fa57" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-28672/natural-quality/output.txt", "bytes": 630, "sha256": "5b3293a881f227bb7b9b9e38222ff87117ee17ba06910d66bcba7e04983d00c4" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-28672/natural-quality/receipt.json", "bytes": 12356, "sha256": "55be5951ed3debe859b4efb4fcd8d0505de1df97dd2e4b04d69eb8b47c17916f" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-8192/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "b433b836c21b6bc2006793482dabaeb8745a1260ef43f6176e1d704d9d81bc85" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-8192/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "d4cd4d68d697b65f295093aab09d6b6b77336293fbf5018270f94c4aaef34e01" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-8192/natural-quality/output.txt", "bytes": 746, "sha256": "d9b24b9b41c42ed16d5882cebb10d901ecc2719d60c81247f59673821a15ecbb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/quality-8192/natural-quality/receipt.json", "bytes": 7718, "sha256": "6e500908234265874aea8400095742c10b7e354aeb55b881ceade403ea7fbcf3" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2589, "sha256": "06b958d09f274b07ce286cec1756fe1f54cb032427015d30d499fc60edd511c2" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/fresh-prompt/output.txt", "bytes": 1293, "sha256": "de2fe9d2152bc1a7f348b2b249207662c7e620777d8ef389da402b66ca6be7cb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/fresh-prompt/receipt.json", "bytes": 10944, "sha256": "8d4120304e43ff88b164f65e038102057d946a09d75920bed03b71518e23c8fc" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "829cf5593c2b8a3b99f3be39322eff3a0ebfd17f24d8342d1d56b35ee445ac03" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2577, "sha256": "2750ee7ae15bdd4560c9147a8bc3fb4f1de1800de72ee79ed401e791528a9a2e" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/reuse-requested/output.txt", "bytes": 1293, "sha256": "de2fe9d2152bc1a7f348b2b249207662c7e620777d8ef389da402b66ca6be7cb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-0/reuse-requested/receipt.json", "bytes": 7082, "sha256": "e6de85b7ed735dc63307c78951e806a1839de1518a2281c8b119d30a51c83f5e" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "7ef5f66a9600bb68903fc75a6b511949acdf509d6eb58ccc250c3eb7e75bcc0d" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2583, "sha256": "ede7c1808b51513ba1c1cff36c73cae59a15e1917fd0b914408030aa658c3344" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/fresh-prompt/output.txt", "bytes": 1293, "sha256": "de2fe9d2152bc1a7f348b2b249207662c7e620777d8ef389da402b66ca6be7cb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/fresh-prompt/receipt.json", "bytes": 10560, "sha256": "fbe7091cd499896801627068a126fcbdd8f08747d947e7a08ecc6f638ecbfee9" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2595, "sha256": "ca6986c3cf98246e029882a0096e476a32526bd1992c901f8aac94ceaac47863" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2583, "sha256": "81e1288c6faf1e84ac4031cfe5803f42adceebd12fff47e2bc1558a2d141cc3c" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/reuse-requested/output.txt", "bytes": 1293, "sha256": "de2fe9d2152bc1a7f348b2b249207662c7e620777d8ef389da402b66ca6be7cb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-1/reuse-requested/receipt.json", "bytes": 7082, "sha256": "6706264e5fa83c34ef5fd746ad9ffc392a51d687ef66dc320f02c795b28377f5" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2595, "sha256": "880ac97936c47955aa1949f18490cddcb9ccb4e35948c977870cea76609c800d" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2583, "sha256": "a7720169030d0ec01ee1d72781d3b76606973019461b7edba84147de398e285a" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/fresh-prompt/output.txt", "bytes": 1293, "sha256": "de2fe9d2152bc1a7f348b2b249207662c7e620777d8ef389da402b66ca6be7cb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/fresh-prompt/receipt.json", "bytes": 10558, "sha256": "8928ddddeb953158d1a380d37c4ba290b0d64a40fdd5c7fa6062bbe87ed44256" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2595, "sha256": "4710bd7b70bc755e331a21449c7097d92b4760f4a1fc74fd31b3d1686498ec48" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2583, "sha256": "49f6edf3d1f70a41ef7349a6bb778cd644c82882c8d6f7d8359f2710e678bece" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/reuse-requested/output.txt", "bytes": 1293, "sha256": "de2fe9d2152bc1a7f348b2b249207662c7e620777d8ef389da402b66ca6be7cb" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/repeat-2/reuse-requested/receipt.json", "bytes": 7086, "sha256": "9021ebccc52111fca73541cd18fef957467002c470d510d645f10738c3052f7a" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/server.log", "bytes": 57765, "sha256": "38c2741b06027d05e6cf8cd00e3c98509b0179d1e4ddbc8cd8f44b40e2ece722" }, { "path": "guided-speed175-native-profile-20261005/batch16384-cache16-t16/speed-input.json", "bytes": 109765, "sha256": "987e187a7f822f558300c4126efd6bce06d5fc62f301d3920fdec62aedf23992" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/cache.json", "bytes": 478, "sha256": "c0ca804e71544276c3284af1c0d4219055b8ae332905fb716447beaa9dafff86" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/configuration.json", "bytes": 1650, "sha256": "fb9c4186871f22d7bce37aad37f53dcfa9fbf0c4bda8f66d124bee4c38e959db" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-16384/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "87b8189928ddc8185ff79d5637110dfee1ed38d2af31f71c6741627a2b0e63c5" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-16384/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "17459b6482432be8d645e0e64fe955fe03e630fa880b32bbd142be8f7a42bd63" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-16384/natural-quality/output.txt", "bytes": 639, "sha256": "632148770e173903ae34adc1439be6d1bccc4369b2d724d623c68680cfbac1b7" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-16384/natural-quality/receipt.json", "bytes": 9240, "sha256": "a8c7fb06b4ec0f4eb5f3135a344ab29aee7f31f5f0e6e70bb015d2a0d33dfaa6" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-28672/natural-quality/metrics-after.txt", "bytes": 2596, "sha256": "49c85f29cd8b53edfcc26a2b105d288eb9902198367bf723daeea39590dcf029" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "9c84c569d63c7c4b9c889d1a752d3ecc92e70385e297d38d89690a02ba2f26ab" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-28672/natural-quality/output.txt", "bytes": 803, "sha256": "5a0a9b2a0095da77ce12223f37c3f5bab27755bfaa6393d6a8c8ed44a71465f7" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-28672/natural-quality/receipt.json", "bytes": 13484, "sha256": "c6ed05ae657c55e2cd2b0a8d0607ee2e363cb43173655cb372c4f92f03ea2e6a" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-8192/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "c5f6a67ca35926e0d941a03fa8b1b3d140de5073fbac3104b3d6553e375fef27" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-8192/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "5585b8faa58141641480986ef2ce2653359116eabfe12ee9652bba0561a38a2c" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-8192/natural-quality/output.txt", "bytes": 721, "sha256": "025800792ce2322d35b6301fea4ecbf635cc1ee1307bfc51f6ef9d2cceb4878b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/quality-8192/natural-quality/receipt.json", "bytes": 7301, "sha256": "2e61abe3e77141df12d13f8ca4708ee5ad123ab9dde2bbb08cc47e4ef1446283" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2589, "sha256": "879d6002fbfcc9c5fa0d55d9ff4f0bc053bbf92e6e453c0af83fd55bac6bdaaa" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/fresh-prompt/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/fresh-prompt/receipt.json", "bytes": 10913, "sha256": "34a926dd6145cdad8f0ee1309369d36a2c3bd8b49ab5ba1eb843b3703b1b0ddb" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2595, "sha256": "a2a5b6aa7e26b45554daf0d68cb7814e3c85e3fa81224b739a6dd24781c447d7" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2577, "sha256": "0e08e6eff157b870d71fb2162f5702a46af93b5e68c0ab05710964c2cbd57db3" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/reuse-requested/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-0/reuse-requested/receipt.json", "bytes": 6667, "sha256": "4bd716ea51543f3369ed031ef31b62a4c318dc7201e5f44a994f3b7716ddbfd0" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2595, "sha256": "3745faa02e608d5353d5a18c7c5341fe11a1656e7e64667c4f1b55a17c19549a" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2583, "sha256": "fb594ab92be6ac389a59432aac92fbbebad2a2aac99b104f0727eb34c692444c" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/fresh-prompt/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/fresh-prompt/receipt.json", "bytes": 10918, "sha256": "4ec3beb753640d2f769bac75d9bd5d817c00817e7af27e45323b46adc3321ae6" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2595, "sha256": "814e5e67f3f928208ee1f49a7e72b5e28edfc3efc116c82e7afa3c39edd31d06" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2583, "sha256": "8250e59bfb7647b29b468dcff0c9a65ea2e31dd0c2e630ffaf91293a724d30ad" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/reuse-requested/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-1/reuse-requested/receipt.json", "bytes": 6668, "sha256": "d9c582d042dbe5e8eff254014ed6ba3f42ecd8dad62fe0a6b9f2b707f9160c23" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "5479d3750baaf275af31f0d146528d7aef23c46becb9254159bdec29d85e34b1" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2583, "sha256": "87049b159784cd510fd4e5218cd55ffd765e5e594515fdd0964a92cadffaf017" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/fresh-prompt/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/fresh-prompt/receipt.json", "bytes": 10916, "sha256": "b233c197fe36b3ff950a7d65e6c890419104a3aad25bd94d76e937ac155449a6" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2595, "sha256": "02512fe9af60a1ab06a27fb97531e5ab1d511393de48735c55f037d7e2583e0e" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "cb4bfb36b97570ff0c8cb3244af59314df9a8bc3e4c0a06ff6a0c9e985859e2f" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/reuse-requested/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/repeat-2/reuse-requested/receipt.json", "bytes": 6668, "sha256": "64a311ecc932d713bf4acde416ebf13a72c57c9415658ef58869c5755ef2d898" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/server.log", "bytes": 60106, "sha256": "eb4cbb27a67ab4faa58146ad16a734a9bc68a904ee2eda1436b845e77d543906" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80/speed-input.json", "bytes": 109765, "sha256": "987e187a7f822f558300c4126efd6bce06d5fc62f301d3920fdec62aedf23992" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/cache.json", "bytes": 478, "sha256": "c0ca804e71544276c3284af1c0d4219055b8ae332905fb716447beaa9dafff86" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/configuration.json", "bytes": 1650, "sha256": "6e1cdb8f05539023e9ae6f1adf3918a4f9dc7dca5470ad551fa13ac88b716853" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-16384/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "36a941c5499273d052bb384b6eb99c49e63e42a9d00be3487d93baa08487dbd5" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-16384/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "1fb60d0c7692117f9912e8e04d2dc343b4a038033b9d7ff262166083ec825b68" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-16384/natural-quality/output.txt", "bytes": 639, "sha256": "632148770e173903ae34adc1439be6d1bccc4369b2d724d623c68680cfbac1b7" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-16384/natural-quality/receipt.json", "bytes": 9261, "sha256": "dade9a83f686378a1aa9d708123a0b6e94d37c9e352202155deaa9b8a869cdd8" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-28672/natural-quality/metrics-after.txt", "bytes": 2596, "sha256": "d939c691dadf00440d0e0da60d01fcb95203f54c5d5ec8e3477a28d7c8e01385" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "f39fe1499d79d988d7e6c15e66579008542dc1d40189565e51b55af05de5b614" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-28672/natural-quality/output.txt", "bytes": 803, "sha256": "5a0a9b2a0095da77ce12223f37c3f5bab27755bfaa6393d6a8c8ed44a71465f7" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-28672/natural-quality/receipt.json", "bytes": 13510, "sha256": "0bf71b829c3cee10baf67971c67fae56240990846ffd0d0a7aeafef2bfbae181" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-8192/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "7c3c01cf025c6b791afb6d9c99d7e304482a3390bf03737c7ae3f9697791881c" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-8192/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "c3542435ceed77457e403c63bdb13728f19cebe6a60215a957295d3f507824b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-8192/natural-quality/output.txt", "bytes": 721, "sha256": "025800792ce2322d35b6301fea4ecbf635cc1ee1307bfc51f6ef9d2cceb4878b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/quality-8192/natural-quality/receipt.json", "bytes": 7330, "sha256": "03886fdceb887d7498ac0527362f42563167adbcd7fcd8b33c115f1818565913" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2589, "sha256": "3a9a1882357991bd7d022b50f51e444a5fe06cf80f95b382d334525c7c28e5ea" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/fresh-prompt/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/fresh-prompt/receipt.json", "bytes": 10935, "sha256": "8c29193dc544ff96a7fe90bac4b768af0b5936513910733cbd0b177f50a08829" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2593, "sha256": "a50120a806f16847a1ab4231850db4bc6fdcaeb8b1578250d9636bc94974e400" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2577, "sha256": "26d329a2ce1edb353afd41a2e41fbe07f8e4c709a86a0ab9fca0d81c5114cbec" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/reuse-requested/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-0/reuse-requested/receipt.json", "bytes": 6691, "sha256": "8d3e332bc0185d7ff6a03fc2e8edb004e0e4c893a09626a2ccdc6e560bfaf605" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2595, "sha256": "601e23665b108bfa45dc86ee89b557a5bf17c22b3ac6b7c5356a549168c5c844" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2583, "sha256": "62b1c6704a621d0dfd19f1427f6d1a3f7fecc7b7577e600341faf4b50c8bb136" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/fresh-prompt/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/fresh-prompt/receipt.json", "bytes": 10944, "sha256": "5bd5a2051b5d5684050541ec6da2a043133f6057d6816f75a656db54037c6dfb" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "193e5f0bbb6baa981b521ebed312ada0da6bb4c704be68753e9f670d7d5351d8" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2583, "sha256": "3fa807c18f693d9fdd81a391e0c860f7bb988b6d6780c4e054cbb3324a8ecca8" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/reuse-requested/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-1/reuse-requested/receipt.json", "bytes": 6695, "sha256": "820e99c79328457fdbf4ae31aa27529b6a902001423786bd9a6f2b4832e6b212" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2595, "sha256": "658d894cef76711c9e0d358034a3085a67113044dcb46ca04e162aad3fa85da1" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "26cc7122596f2095e9306b6af7d8e746f1e3643403ad79fb10113cc1a75dc8f4" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/fresh-prompt/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/fresh-prompt/payload.json", "bytes": 142785, "sha256": "f96a1bf62952cbdb4418ee2aed35294a61c5d9a2ccfba22b583860c36c75c5b2" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/fresh-prompt/receipt.json", "bytes": 10943, "sha256": "ae4cf839b7fd2ea0e83544feccdfbe6c01b3a34d7d27061a7a2626c67297c8a0" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2595, "sha256": "7fa5120e46fc0ff44d930c3023416294932cb160518f728c25ed13bd42226fe9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2583, "sha256": "1293c3b695e199556fafb1680a880d473b492458de52e0cc0c37ff19554141d6" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/reuse-requested/output.txt", "bytes": 1301, "sha256": "16e90ee623e1b87dc23345b3ca37f61be55a57ad38624a520fcdca1ede846a4b" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/reuse-requested/payload.json", "bytes": 142784, "sha256": "a68a08fcfc420f700c806ec6d0d6429dd4a2eeff1790064719b8eef1556b16a9" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/repeat-2/reuse-requested/receipt.json", "bytes": 6693, "sha256": "bef68f1df66fa47878c8a1c7a1368bee16e18d31bd13bcfcdc58af58443bace5" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/server.log", "bytes": 60106, "sha256": "acd779e4a53f818214a152ca2ee0f51c13e939353670dce6bb32bd30818ccfc3" }, { "path": "guided-speed175-native-profile-20261005/batch8192-cache80-t16/speed-input.json", "bytes": 109765, "sha256": "987e187a7f822f558300c4126efd6bce06d5fc62f301d3920fdec62aedf23992" }, { "path": "guided-speed175-native-profile-20261005/requests.csv", "bytes": 3480, "sha256": "0d595e3b8ab8ba645b6d23bec196ae621847eded458f5f8192737897c7a65a2f" }, { "path": "guided-speed175-native-profile-20261005/requests.json", "bytes": 262865, "sha256": "a5de9c350e3f38db3a515822d286dac4b807d4f375658249016b97c3aa1127ab" }, { "path": "guided-speed175-native-profile-20261005/resources.jsonl", "bytes": 1121408, "sha256": "c31a69bf29a6f12cbdf7d3a940dee07f14a4a36ce7d6a470a966ad9e88340144" }, { "path": "guided-speed175-native-profile-20261005/summary.json", "bytes": 11914, "sha256": "e123af9cdcca4346e839c21d3b30755f9087429d68f54124190ca3ac28fd3c20" }, { "path": "historical180/mtp-screen32-20261005/control/configuration.json", "bytes": 1795, "sha256": "b15bb1ecb94ff5b10bd8fad553f478c17de1775ef37f7ef94af1cff4c04fb150" }, { "path": "historical180/mtp-screen32-20261005/control/physical-server.json", "bytes": 344, "sha256": "554799e75a180283a58c397736f1713be310fa860e87529b3575434f46844ac4" }, { "path": "historical180/mtp-screen32-20261005/control/server.log", "bytes": 1722, "sha256": "e4d1a0a90327742984ae8ffca9479580d7030d8e5f49b5d088f692fa02fcc940" }, { "path": "historical180/mtp-screen32-20261005/requests.csv", "bytes": 286, "sha256": "3fa10d367ae89bc81e38726a08e7e0e456b7fa1e84dad9950b1ac0b31c09dc4a" }, { "path": "historical180/mtp-screen32-20261005/requests.jsonl", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/mtp-screen32-20261005/resources.jsonl", "bytes": 92899, "sha256": "b46ffde03834e00b7addababcd3a917e2c0eedf0653edf5217fdd5cdfc0e1b09" }, { "path": "historical180/mtp-screen32-20261005/summary.json", "bytes": 24730, "sha256": "c11095ce7e544e9351ddffee024a3262fec8b7cbabca65bfa0f88cbb23a5d8bf" }, { "path": "http-baseline6-cache32-vision-20261004/configuration.json", "bytes": 1424, "sha256": "8c8829a8ee0b31c41962533e19d9b0a63b15fdba5e9d67fe24a53c1eb3236f03" }, { "path": "http-baseline6-cache32-vision-20261004/context-131072.csv", "bytes": 2040, "sha256": "53a60e7648dbaef9187b02081c139f60826eeacd133b2a330ba17dd2119064fa" }, { "path": "http-baseline6-cache32-vision-20261004/context-131072.jsonl", "bytes": 99068, "sha256": "c12b7ce150a20b18751f7f28a33b68e621cea7849236d0f9fa15182c6544aebc" }, { "path": "http-baseline6-cache32-vision-20261004/quality-131072-8192-0.txt", "bytes": 567, "sha256": "4d36b710def51e070a5883315bb989fbec1d25216675f363cc69ba3717e041c0" }, { "path": "http-baseline6-cache32-vision-20261004/quality-131072-8192-1.txt", "bytes": 528, "sha256": "e8d088cf34204a840504b084948556894a09e6bc25132333f1e2783002cb3d7b" }, { "path": "http-baseline6-cache32-vision-20261004/quality-131072-8192-2.txt", "bytes": 615, "sha256": "4fabc418f42451293fb08483645dfc019ee9ae95a6ac5d91080126be3b28f9c2" }, { "path": "http-batch1024-cache32-vision-20261005/configuration.json", "bytes": 3691, "sha256": "3d371913a816640ec1877fc7cbdc45bc7460396dbb7a37f4be456c5518e7fd2b" }, { "path": "http-batch1024-cache32-vision-20261005/context-131072.csv", "bytes": 1520, "sha256": "d91cf29c750ef555d8a8503cbaa435880327541bb732ac17ec7a66d70cf2857c" }, { "path": "http-batch1024-cache32-vision-20261005/context-131072.jsonl", "bytes": 50841, "sha256": "1b0a5a4038d1552be49d49e156116f6c45f31d221de71f0239119fd135b4b42a" }, { "path": "http-context32-cache32-u1024-20261005/configuration.json", "bytes": 3628, "sha256": "53b018526e9b800739dce80320aab9d1f4873f549440c1609ff274e79c6756cd" }, { "path": "http-context32-cache32-u1024-20261005/context-32768.csv", "bytes": 1491, "sha256": "cec2ed09c71473d0eb14f42e9462944ae0cffab4e29338d1ba65642d129a23ed" }, { "path": "http-context32-cache32-u1024-20261005/context-32768.jsonl", "bytes": 50528, "sha256": "695653049a68a6bc477d1cc8e1cbb36daaa6a8a53c8c3869a3750a44e7d5104f" }, { "path": "http-threads14-cache32-vision-20261004/configuration.json", "bytes": 1389, "sha256": "f2a316eff1d06613abb33796f06a6809a7fe92b1fe3a51c45acee4758447681e" }, { "path": "http-threads14-cache32-vision-20261004/context-131072.csv", "bytes": 1480, "sha256": "cd7d18b930772cd59f29b8bd77bfdb089acf2ce42e84e615d4e1804bd49e6481" }, { "path": "http-threads14-cache32-vision-20261004/context-131072.jsonl", "bytes": 60406, "sha256": "0f0a0aac41c496113da8310148783650f2a4c7f12c3a8701cf5bd3c8e8380f53" }, { "path": "ik178/results/control32-noMTP/quality-28672-0/output.txt", "bytes": 590, "sha256": "cfde1ae6628f58d89ed448790e0273cc59996e6fe68cf854b62e51faf3f9a6c1" }, { "path": "ik178/results/control32-noMTP/quality-28672-0/payload.json", "bytes": 249525, "sha256": "3cba61a6792cf8982286162eb252d55c2bb51c905e30dcbfa7622c3169c3c917" }, { "path": "ik178/results/control32-noMTP/quality-28672-0/receipt.json", "bytes": 9461, "sha256": "c9ff37633e5df7bd40c791d51a99440f08de34d3cab52441f6d3f2796e6f5d47" }, { "path": "ik178/results/control32-noMTP/quality-8192-0/output.txt", "bytes": 894, "sha256": "1087ac514d351150e76dd18149d210b64b958a1adbb49d3c12aa502406bf14f8" }, { "path": "ik178/results/control32-noMTP/quality-8192-0/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "ik178/results/control32-noMTP/quality-8192-0/receipt.json", "bytes": 6356, "sha256": "b5e994f5f4b7822127ed09ee15e76e7f6e7fba528f960237e579fe51e64169b8" }, { "path": "ik178/results/control32-noMTP/requests.csv", "bytes": 747, "sha256": "1645062eafbe581a31e6a36f7bdd0051df7340762b2aa0af73b5788e16924e73" }, { "path": "ik178/results/control32-noMTP/requests.json", "bytes": 36758, "sha256": "ca40d5e5b17d4f6f759460c4acdac5fbd606fb28cbbd3acb8284d11c7053356e" }, { "path": "ik178/results/control32-noMTP/resources.jsonl", "bytes": 278267, "sha256": "cddf9c9b8a8ebc72839737b3f7b97a39bf7c8ff7a50dc16dd0b6c7522980f534" }, { "path": "ik178/results/control32-noMTP/server.log", "bytes": 49498, "sha256": "4d0a3706d34c2c78c885e62662075582fcce54f344e5da7da4d48c1c53ae2b34" }, { "path": "ik178/results/control32-noMTP/speed-28672-0/output.txt", "bytes": 1216, "sha256": "94557c9c2c612298f56cd2a9a119f3fad5af003a0ac823763230a0269e950756" }, { "path": "ik178/results/control32-noMTP/speed-28672-0/payload.json", "bytes": 249557, "sha256": "19766cf4c14b17729e8087d3260c4a66034d819410f62423c9f721a6af965b3b" }, { "path": "ik178/results/control32-noMTP/speed-28672-0/receipt.json", "bytes": 11144, "sha256": "43dc007725aaea347fa1a82637a269f97b73567b0db6c94b018a1e4816fc85d9" }, { "path": "ik178/results/control32-noMTP/speed-8192-0/output.txt", "bytes": 1076, "sha256": "fbfed364d63aee5b0257a17c7d8d64e44e7b296aeb2e8b3c744f8006676d5cb6" }, { "path": "ik178/results/control32-noMTP/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "ik178/results/control32-noMTP/speed-8192-0/receipt.json", "bytes": 7266, "sha256": "399f6d5bba9a408c5ff820b4d931e68f5864fcaac799956e6ceb87c1fa1d6961" }, { "path": "ik178/results/control32-noMTP/summary.json", "bytes": 5586, "sha256": "9b9be58d9282469e9c2aba8149c325b6a58226fe8d4ed47a845edb502d3be76c" }, { "path": "ik178/results/mtp32-n1/requests.csv", "bytes": 146, "sha256": "eb9e980aac7ace44eefe900bca780d2dc89afb76f9d4037baf7df4fd0ed1161f" }, { "path": "ik178/results/mtp32-n1/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, { "path": "ik178/results/mtp32-n1/resources.jsonl", "bytes": 111182, "sha256": "7fb3019f847b1b898b1a4a8fd74243544b9c49ae11685e3b2c35c50aebe9d989" }, { "path": "ik178/results/mtp32-n1/server.log", "bytes": 44392, "sha256": "39a9faa694d7b04dd6febf3722475e4183fdf0efebc6d6be3281743bda45e241" }, { "path": "ik178/results/mtp32-n1/summary.json", "bytes": 5169, "sha256": "3a806194e6fb5f40482c99875e9d4515b995d654195f487ffff2c8e3c82fedba" }, { "path": "ik178/results/mtp32-n1-draftU512/requests.csv", "bytes": 146, "sha256": "eb9e980aac7ace44eefe900bca780d2dc89afb76f9d4037baf7df4fd0ed1161f" }, { "path": "ik178/results/mtp32-n1-draftU512/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, { "path": "ik178/results/mtp32-n1-draftU512/resources.jsonl", "bytes": 116573, "sha256": "00090327a96ac6ae2f469d5a7faf438606a515afdc07f995f2ef9fb87e615899" }, { "path": "ik178/results/mtp32-n1-draftU512/server.log", "bytes": 47191, "sha256": "2ed71339505bb236455c2248ea08fb2c41787edf0540df6900625d75e005ba99" }, { "path": "ik178/results/mtp32-n1-draftU512/summary.json", "bytes": 5590, "sha256": "6136b95fc0c866ba4e61fbf3487405d6ee451b17ea050a65a16c10a5ddcc7bcc" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/quality-28672-0/output.txt", "bytes": 703, "sha256": "6928b8bb3c552dd91e923c77397d5b6a93a5496ba6f6cbd0baac2c1319555548" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/quality-28672-0/payload.json", "bytes": 249525, "sha256": "3cba61a6792cf8982286162eb252d55c2bb51c905e30dcbfa7622c3169c3c917" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/quality-28672-0/receipt.json", "bytes": 11599, "sha256": "76cc60faa5e4c08aa22541c9ea9e9c6b340dc569ef9ed9980da1d9db52c529d5" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/quality-8192-0/output.txt", "bytes": 697, "sha256": "9f4bd337a5ae5265247a48192fad8128955188d1e28af992c60c7f1e82fe5a89" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/quality-8192-0/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/quality-8192-0/receipt.json", "bytes": 6555, "sha256": "89c796a4e7bc3da8e92fe5d12d7768481d1a91084e0caed7dc878f97ce0c62a2" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.csv", "bytes": 738, "sha256": "a1ae5d59b778e5017e1e67e0cac9afdb3798c7ef59b5d5ced056096c1ff0548b" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/requests.json", "bytes": 40991, "sha256": "b0c4b235b06560ccf360509232909a39c3a61c50623d801034a31ca09e12f274" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/resources.jsonl", "bytes": 297125, "sha256": "893592c49ce065b3dde94ea88ca6bced8fa68ddc23b8594b81e184b305fa82e5" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/server.log", "bytes": 64287, "sha256": "946019e5d89269a24a6d6f67acb32b0b17b193d83f156a179309478275e1f54d" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/speed-28672-0/output.txt", "bytes": 1175, "sha256": "a98c9845a27ca785de8b5a63d85112b0acec5cc03ed2e906ff0b67a9081b6ff0" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/speed-28672-0/payload.json", "bytes": 249557, "sha256": "19766cf4c14b17729e8087d3260c4a66034d819410f62423c9f721a6af965b3b" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/speed-28672-0/receipt.json", "bytes": 12512, "sha256": "19ce78a761f7f2e6d11ca4e654fffbe20090e4adebbbe67c59ea3970b7a2a376" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/speed-8192-0/output.txt", "bytes": 1506, "sha256": "484e33a6fd5effd5566e70a15db56e2bcff8a235ba7ca85b44cdb02ab9933ee9" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/speed-8192-0/receipt.json", "bytes": 7474, "sha256": "67354c52b31700a45292d6b3b41656f663c1d0d8826789c09729bd3313b1c191" }, { "path": "ik178/results/mtp32-n1-mainU4096-draftU512/summary.json", "bytes": 6255, "sha256": "37c8d10eb66e50bd0774a19f93ab523e39249403d4c38acdafd0988ebff3986d" }, { "path": "iq3-author32-accept-20261005/author-control/cache.json", "bytes": 369, "sha256": "1b2b9e7831548c6bc9ca834ef9f81ce1033ff113aaba2370f0b0e0c8483b593d" }, { "path": "iq3-author32-accept-20261005/author-control/configuration.json", "bytes": 1594, "sha256": "3e804083f04371e0fd259245f4c8e7327d54917604cdd3d8237aaa59d90d84f4" }, { "path": "iq3-author32-accept-20261005/author-control/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "iq3-author32-accept-20261005/author-control/quality-16384/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "f06b86d9eb3ea7be54d56e4c22e81f200bdee745206de83331ee181736e799c3" }, { "path": "iq3-author32-accept-20261005/author-control/quality-16384/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "8006120c6f5fcaf97f5a8291b02a4bcac2b7d1a8e3af6eb2a33f930dec4e7054" }, { "path": "iq3-author32-accept-20261005/author-control/quality-16384/natural-quality/output.txt", "bytes": 603, "sha256": "3602e87178013c3996a9397d3362aa411dbc2b45b5e50dfe2e9b192677cad068" }, { "path": "iq3-author32-accept-20261005/author-control/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "iq3-author32-accept-20261005/author-control/quality-16384/natural-quality/receipt.json", "bytes": 22670, "sha256": "78e36f836599eb1205f68c6d77582737d50e1803fce6272d6c08c5902bd8fd2d" }, { "path": "iq3-author32-accept-20261005/author-control/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "iq3-author32-accept-20261005/author-control/quality-8192/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "aa66e5504a1ca033eb4f7baf4e969f304ff623a9633866003d8696f569a95ae2" }, { "path": "iq3-author32-accept-20261005/author-control/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "8233e3514afcc865247fd11349c0aa0ab2d29654b082b4e969562e64220aa793" }, { "path": "iq3-author32-accept-20261005/author-control/quality-8192/natural-quality/output.txt", "bytes": 644, "sha256": "e9db1312739f5709f5c5a56e04097881e218e35d65fc541c18988a4ec3eefe8c" }, { "path": "iq3-author32-accept-20261005/author-control/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "iq3-author32-accept-20261005/author-control/quality-8192/natural-quality/receipt.json", "bytes": 13788, "sha256": "4de4bcc0f43984c7d3c7b99c1ab87d4b25e3cad4225475f3995f9455ee99a467" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "a1ddbbcdc0b9d35eb9892127e5a405b270c8468053f62a3781cd44055624753d" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/fresh-prompt/output.txt", "bytes": 1224, "sha256": "05095281fb348d40d603b7e20e463748bc7306ae731f31c5426564636a3bad38" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/fresh-prompt/receipt.json", "bytes": 17387, "sha256": "2b8ae928404fb598a58d3a7c2879b5756e20480fb226acff14ab2daf4d85cf61" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2591, "sha256": "1e4b2cba90cf3daf40c11afc60f169459432d6485fffa7003d9cf83af479cbbf" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "c9468fc4d67084a71a557393c4a69cfbf1ae48c78cefb8d363de86a36b824b65" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/reuse-requested/output.txt", "bytes": 1224, "sha256": "05095281fb348d40d603b7e20e463748bc7306ae731f31c5426564636a3bad38" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-0/reuse-requested/receipt.json", "bytes": 6976, "sha256": "5e559d5b00552a9987a4071336e322afb2cdc2905201a283656e832016ea0640" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "8b4a26c58b19e6d1bed4300ab180ffbfd82f701839ff28d975f740725709ac80" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "aad7c6a89f3055df448313d0eb26e4e96769f8e4e452817d088f6a9b43134265" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/fresh-prompt/output.txt", "bytes": 1224, "sha256": "05095281fb348d40d603b7e20e463748bc7306ae731f31c5426564636a3bad38" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/fresh-prompt/receipt.json", "bytes": 15467, "sha256": "768a4a2b44481f3577a35ab3d9e5a1dd12e1743d50c310b585c149ba5872d214" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "11e14206e343a83f56ff196094d71448cfca39c00e6d481641c5044febfe3a8e" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "0ce0ccbc21091bfafbc8912cb95e40a5e397443f5d67f5b749ae6007ff22e5ce" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/reuse-requested/output.txt", "bytes": 1224, "sha256": "05095281fb348d40d603b7e20e463748bc7306ae731f31c5426564636a3bad38" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-1/reuse-requested/receipt.json", "bytes": 6970, "sha256": "5d5ca192100934ff4b6e05ed43d20a56f00c5e7dea580afcddbd23941d9f0aa6" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "593227dee71796a796e667be4a02d730eea1bf5cab9ef49c585fbe2b8761a99d" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "c024d7a4037cbdd4677351eedf26c29903888a524ba21ac6604cc65e9b77902c" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/fresh-prompt/output.txt", "bytes": 1224, "sha256": "05095281fb348d40d603b7e20e463748bc7306ae731f31c5426564636a3bad38" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/fresh-prompt/receipt.json", "bytes": 15464, "sha256": "63c422603d49bb5ca87e2b21463717e4f182c685a083a9bddc0b3c0ae6de053d" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "017063b60630dc70c11d7a9e4e2cc5b79fa712b4d2a35003e5cad0c76d7d552d" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "3a0604cb2d1ab3e5f41651ad89fb609b4cfcfc6e8eed620f0edba0c3b25a0c31" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/reuse-requested/output.txt", "bytes": 1224, "sha256": "05095281fb348d40d603b7e20e463748bc7306ae731f31c5426564636a3bad38" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-author32-accept-20261005/author-control/repeat-2/reuse-requested/receipt.json", "bytes": 6972, "sha256": "5f48f53f2e612e95e6e48acce017e8b52709f094818431368208c1fdfee3fee9" }, { "path": "iq3-author32-accept-20261005/author-control/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "iq3-author32-accept-20261005/author-control/server.log", "bytes": 60558, "sha256": "a10736487123512ca16c8a5db181982decf37fbced3b3298e99dd8712f1da28c" }, { "path": "iq3-author32-accept-20261005/author-control/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/cache.json", "bytes": 468, "sha256": "822c78d79af5d5fc7a068a0c0c838482c48fa2a80737c25290fd80e96e4e88e5" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/configuration.json", "bytes": 1919, "sha256": "c5173776d3d6a455bd55c3b9bdc541505bbd39a1e90a6da1832a17183ce1e487" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-16384/natural-quality/metrics-after.txt", "bytes": 2844, "sha256": "a2db747e09adbd7060579aa279833f9e1c042ba0f0c9c032ffd2373cf3f60ac5" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-16384/natural-quality/metrics-before.txt", "bytes": 2832, "sha256": "25675a2a91a6d481af8ea04976201830e24deaa6432caade03c4356f2c333ee5" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-16384/natural-quality/output.txt", "bytes": 560, "sha256": "bac132bf776759345b09b8213d964d9d946c369f4358bf5bbadb18c253c83a84" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-16384/natural-quality/receipt.json", "bytes": 24197, "sha256": "70ccaaffd60d794c2c2ff77305cf98a118c31c5358151430cb307695d5cbe95c" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-8192/natural-quality/metrics-after.txt", "bytes": 2844, "sha256": "91828905f22d191d918b83dc07c18e41b93d11640c243d5f9c2bf27254359c89" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-8192/natural-quality/metrics-before.txt", "bytes": 2832, "sha256": "2f740ea97ce9afc06c389af854089afda001ac49bcf99d45d0bd7dbece3eaa75" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-8192/natural-quality/output.txt", "bytes": 700, "sha256": "b6af93ff338c6992e83ce191d81c75bdbfbfb6e4690b30bc4f706349fe90340a" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/quality-8192/natural-quality/receipt.json", "bytes": 14546, "sha256": "eaa2972f25af71f6f38f4e0a8362f038f8122def6adc2e91210e2882c6a9e7e4" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2833, "sha256": "c5155caa9690ea6f999727c0f132fd96a57065736b9774ac9444d66363b4d5b5" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/fresh-prompt/output.txt", "bytes": 1208, "sha256": "ee7b41c7a37ee139b0c4a816235f1b8b0fdb6649036592d2c43bb9c6738f8e63" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/fresh-prompt/receipt.json", "bytes": 16611, "sha256": "0b2dc4f1e4e7f0fa38e2bf3e633421df6945c253ea12c6f63958b42f72f0607a" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2836, "sha256": "2bfaf257e3ef9a054a632d3121d8f63adb8c23374a53c262895edbbb86ca7897" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2821, "sha256": "d9d844522c1ff4dabab29ea2dd818a1e16a4a4e1e28a2455a837cefd84ff896f" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/reuse-requested/output.txt", "bytes": 1208, "sha256": "ee7b41c7a37ee139b0c4a816235f1b8b0fdb6649036592d2c43bb9c6738f8e63" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-0/reuse-requested/receipt.json", "bytes": 7733, "sha256": "3c7dec29d24bdba8200621f46ea12fca65c8d09342f5315245c585b7cdb833b2" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2836, "sha256": "eafb2e3f4815075cf7391100517f1798c59cbde32e9203db1b4c4120bd59efd9" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2825, "sha256": "1f93b573751fc029adaa6fe575d6d3bdcd2abbd6233e4bb86645f075c38d72b9" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/fresh-prompt/output.txt", "bytes": 1208, "sha256": "ee7b41c7a37ee139b0c4a816235f1b8b0fdb6649036592d2c43bb9c6738f8e63" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/fresh-prompt/receipt.json", "bytes": 16609, "sha256": "e3af190c3f2a30358ec9d7c6178d940f69874225ebe3a46e08c03e5db60f11c9" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2841, "sha256": "7463f4e7a34992b6babc570415a899754dfcd77a47eb28eb3bdff8e0ff31328e" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2826, "sha256": "30c2f4d12a5efb52ece108e572da1fd3a234610bac89bac414ac27171e632809" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/reuse-requested/output.txt", "bytes": 1208, "sha256": "ee7b41c7a37ee139b0c4a816235f1b8b0fdb6649036592d2c43bb9c6738f8e63" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-1/reuse-requested/receipt.json", "bytes": 7734, "sha256": "d89b9c15048013dc12483970c405ce6a70a3c1067c03fb3909eabba4a45c11a1" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2844, "sha256": "fa46df805a42fc94f3e81451f1c27a144c458967781bec769a578a17f7445a68" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2829, "sha256": "eb3c279611f9aa95392d8a061caf5f67bcde335c4b2a32d00e982542bdba8587" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/fresh-prompt/output.txt", "bytes": 1208, "sha256": "ee7b41c7a37ee139b0c4a816235f1b8b0fdb6649036592d2c43bb9c6738f8e63" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/fresh-prompt/receipt.json", "bytes": 16610, "sha256": "d853c88771917d4788a2dff02b17f9ece4fe07c26d23bacc08dadc1af0670b40" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2844, "sha256": "c87673710df235556377d8032c13d955d47a3c26b1cb142fe346120caab08c04" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2832, "sha256": "639afd7bce46e7c0ea7264ab1be93a9a4a759a97ea06a5160d8fbfa7ac0af406" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/reuse-requested/output.txt", "bytes": 1208, "sha256": "ee7b41c7a37ee139b0c4a816235f1b8b0fdb6649036592d2c43bb9c6738f8e63" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/repeat-2/reuse-requested/receipt.json", "bytes": 7732, "sha256": "a8495eecbe5276c9a252c9d1982a4c243b8231eb5d985868468ca2a5c8822cd8" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/server.log", "bytes": 78866, "sha256": "f92d7af9fd414690f2b5ea0b80d8127eb5c91dd3a85f81b393e347c819dfc73d" }, { "path": "iq3-author32-accept-20261005/author-cpu-mtp1/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "iq3-author32-accept-20261005/requests.csv", "bytes": 2135, "sha256": "407b7631299e4e12c1d01877ca7eca162d31102a74e2be903535b40ba11d121f" }, { "path": "iq3-author32-accept-20261005/requests.json", "bytes": 233253, "sha256": "f3f9fcf1cb446ae9d3d66e67c9451c11d36837e1ce0ea2161fec967366da7f9a" }, { "path": "iq3-author32-accept-20261005/resources.jsonl", "bytes": 1204121, "sha256": "223f6467c44e694c1331e61f2a9d4eac780f3f0a40029e3571015ddffd01714b" }, { "path": "iq3-author32-accept-20261005/summary.json", "bytes": 9686, "sha256": "194b8d985220e729963077558d0fc8c195ce7b088bad39f6a6bfb5b8360cbf5d" }, { "path": "iq3-cache96-accept-20261005/cache96-control/cache.json", "bytes": 369, "sha256": "7419b934210768c20c567c2beeb09d0e2a1c5567f9c603fcdb2de958b802c6de" }, { "path": "iq3-cache96-accept-20261005/cache96-control/configuration.json", "bytes": 1594, "sha256": "9d078f937789dd7fcbacb68dcd9e5be6b4c9b1d57ed111bce03e2f8561eef3b8" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-16384/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "cf7970f0b56d8d07512d50f92d29e0170bed5f6cc8d49f21c3c45d8950f4b9fd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-16384/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "d79f1f56cf436ab0171f4b11110bc00dd82f08332cbbc21ea5c5973167ac8031" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-16384/natural-quality/output.txt", "bytes": 555, "sha256": "c0c485591944f4af3405f27269919a4089b14ca7008a9d0741a8079af9bdb0b4" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-16384/natural-quality/receipt.json", "bytes": 21507, "sha256": "89d4fe2cfa5103387cdad3f6a1cab0fa8a51fa067eb9ff6c657835af46b1bd51" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-8192/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "32e4864f59dfcf8c0164878db23eb7353c923117776c7dfea2a518aa86e0a5e0" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "c471aece2dac386565fdc46b62b96912870029db4c7ea219fb0e9aab21160198" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-8192/natural-quality/output.txt", "bytes": 857, "sha256": "eea85a20ff50fc6a53a08fad2839d58dd531619edd662c8bee6294b26f613c19" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "iq3-cache96-accept-20261005/cache96-control/quality-8192/natural-quality/receipt.json", "bytes": 13400, "sha256": "757dcd1f036ec7f468e53909b8def7c40633c9afdf7c67ecbaf6e45962346349" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2586, "sha256": "7ec7f1d28f294f1650966bf348e7990a0c0dd9a9cbbaad7baa468eabb63a575a" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/fresh-prompt/output.txt", "bytes": 1569, "sha256": "ec369b6ed21d160669210c631483cf18c783c66833a31c793beff5585cc6a8bd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/fresh-prompt/receipt.json", "bytes": 14692, "sha256": "73e760c1a72d7ae2418c7deba3b7c5a2f473ad639ca325f6753232a538b75e72" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "275cdd80680146d08a05ae3dc59c6b2bb33d9a59b8f7c09018f4e47a7e549f4a" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "f88dbabe2a9bd920cb7676ea020535683df4157c9b25a7c25d32358da212f2a9" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/reuse-requested/output.txt", "bytes": 1569, "sha256": "ec369b6ed21d160669210c631483cf18c783c66833a31c793beff5585cc6a8bd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-0/reuse-requested/receipt.json", "bytes": 6591, "sha256": "405b3d2ffe5dc0f89178e8575c7b01faa53dc46159831efe05734570de99fea9" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "7e2df9d03b9ebbb5685a2472230e91960c9362f28432ca34135e632cd353280a" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "22e8223ddb6dea11442131eaf6d5ac2b2e6f996eb6877aeb4954519e73880556" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/fresh-prompt/output.txt", "bytes": 1569, "sha256": "ec369b6ed21d160669210c631483cf18c783c66833a31c793beff5585cc6a8bd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/fresh-prompt/receipt.json", "bytes": 14697, "sha256": "9774a8648361eda1a542d32610dccee3a192d310a8081cca560370478f27213e" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "292591e5003a0c928135192791e0be512f6a0657a7201e7a0b2a4f076bb8a59d" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "e589ac6ce5c9f027fa12d0f9e861d5103745896e8c57e84457b8f5ed7d85dba0" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/reuse-requested/output.txt", "bytes": 1569, "sha256": "ec369b6ed21d160669210c631483cf18c783c66833a31c793beff5585cc6a8bd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-1/reuse-requested/receipt.json", "bytes": 6589, "sha256": "5bf0f8dae533fc7df4188788ebdf37ac99eafd53f9cf4bad458b89c533dfa5b2" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "15eb2bbf4d9707ddc98185a5c0b3ae23ffde906a4e3aa373583074422de6d99e" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "f070258d9a113c41826397c24de16de76f506e55e8ebaf216eb96045cfdf839a" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/fresh-prompt/output.txt", "bytes": 1569, "sha256": "ec369b6ed21d160669210c631483cf18c783c66833a31c793beff5585cc6a8bd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/fresh-prompt/receipt.json", "bytes": 14697, "sha256": "3aaeb69780465c08c157347e786b74619b65c513afcd5d211122d1b670133261" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "a30464ae4da03ef92cfb76bb6649573034819cfe58b140f12439db70ac67ecfc" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "d21d8920cbccdc5f77d316c179e32bec1745ca4c6f11d05cb32c43f80cda430c" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/reuse-requested/output.txt", "bytes": 1569, "sha256": "ec369b6ed21d160669210c631483cf18c783c66833a31c793beff5585cc6a8bd" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-cache96-accept-20261005/cache96-control/repeat-2/reuse-requested/receipt.json", "bytes": 6585, "sha256": "ca7f73955e91ff35d7dba1bb7a3eabf3603693e6cbc17edc25f9ab1adf64f33b" }, { "path": "iq3-cache96-accept-20261005/cache96-control/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "iq3-cache96-accept-20261005/cache96-control/server.log", "bytes": 59923, "sha256": "2afbf14684c8a0d0098b9c6c62a5dd075475a8806cb6ea4c8f23e1651ea035cb" }, { "path": "iq3-cache96-accept-20261005/cache96-control/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "iq3-cache96-accept-20261005/requests.csv", "bytes": 1056, "sha256": "668e82011c2907793cbbf90a968dce2d211765f120e6bbc253685fd2791d0f9c" }, { "path": "iq3-cache96-accept-20261005/requests.json", "bytes": 105937, "sha256": "2465744e3a014cdb585b0feb8a3483104a7ec5bf102117184125a63da638830a" }, { "path": "iq3-cache96-accept-20261005/resources.jsonl", "bytes": 567977, "sha256": "7af9d76c6653d2ec0018e7181a65a48c2e9c190631fae74165618ededb66b7a6" }, { "path": "iq3-cache96-accept-20261005/summary.json", "bytes": 7359, "sha256": "83c61070b3e2220f45254a8aac6a57b4a7b3ec6fb22a7b6f3261ae45df3f3399" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/cache.json", "bytes": 369, "sha256": "7419b934210768c20c567c2beeb09d0e2a1c5567f9c603fcdb2de958b802c6de" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/configuration.json", "bytes": 1597, "sha256": "e388672909f4b19d48eabc736a2a03fee5bf24de9684c83e019b47db4a259d59" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-16384/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "2eff6c70cbff6d9b296e5d4f692610f75c31fe9bcc7a1eff3850e4a0afdc4cc3" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-16384/natural-quality/metrics-before.txt", "bytes": 2581, "sha256": "05db2d2e1a92ce40f0ce5ecddcf7a9e93a2cfdd1ec90c640b95227af3a2214ed" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-16384/natural-quality/output.txt", "bytes": 736, "sha256": "bbc8b7dee9252ce3f91967fc715bb9328d6d3ee4da70685155e8a9ba2cdec44d" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-16384/natural-quality/receipt.json", "bytes": 10355, "sha256": "e1bdec1de6fb00ea19dc1c0cf6fc45785dc57dac6493086174325aba79d6b21d" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-8192/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "ade1a5966c1ca2a4c07634cf71080fbce75b69cfbf2a867afe18085eb9691134" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-8192/natural-quality/metrics-before.txt", "bytes": 2581, "sha256": "f021b7fac5c4f4f2bd7e0f4c8e4a286d6220e8dfe354c276e3c47b55c5eddde8" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-8192/natural-quality/output.txt", "bytes": 569, "sha256": "5f01f8b922e954c482409991ba4ea87ca62a52bdd6ccb8d1ac78ca3613c4378c" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/quality-8192/natural-quality/receipt.json", "bytes": 7263, "sha256": "59197a94fc9e2311dc5f579571bfa5f3c8f6bb9a15bc25354690ad7c382d4583" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "fe850b24a2a85b934bcf318f9b7bf8fe582d7ce12411ca0f9ef37a7b86d27b86" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/fresh-prompt/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/fresh-prompt/receipt.json", "bytes": 9327, "sha256": "57f3f0ec187e838b265d62a377505e9dad3dec31048d62c654a135eef977c4ef" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "347258ae4b5edafef4f68da805a4f4c2bf7736fe14ca23f588253974dfab0a78" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "246bffc34ed897a8665c09c337471f9d44d28afbb881c9a4348e6ff4e1a31757" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/reuse-requested/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-0/reuse-requested/receipt.json", "bytes": 6622, "sha256": "ddf43b2890bca37d057e24384cd9b527fbe50b49537833a45a30b70cc5d88aae" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2592, "sha256": "3999c41dbf9003778cc1c2f15e23276ea454c30cdb6f58a5e275cbffd021446e" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "0aa65684b2144639e1357d16f903620bcde71dae1ce3fe05c656b8bfc855191a" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/fresh-prompt/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/fresh-prompt/receipt.json", "bytes": 8942, "sha256": "18fedf839c52e7329a3faef7a4a232128b04c5b924dbae6dec87163c844f9ffb" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "fb2ec222299c7cbd2f94bfbaf6c96d3ed035c00891844ca1803fe3a3699128c0" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2580, "sha256": "0fa9b437969cf3f70c58c89597f7cdc0de7d9f1428691d1f0b479fdc83799c06" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/reuse-requested/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-1/reuse-requested/receipt.json", "bytes": 6234, "sha256": "c1a781243a23c1ae3f51c31db9e0270f925e8a2012e52b7ebd19cb01b0548934" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "cd8a85f56cc17a9932aebd169cd9f2f40a142d411a9ba2f9f5535d4d9165beec" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "d5d27254a611c0a99a4116e0883f0a85828849545d7f5c936e571c31498e4916" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/fresh-prompt/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/fresh-prompt/receipt.json", "bytes": 8945, "sha256": "8eb4ad36b337fd2ecc5c132539d3c7950df02f1d2bdfc54201713e806501ba64" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2593, "sha256": "0a32b07bce836f80c33548c92104c9db1e67f966af4e4c2bd3959a1710047b50" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "0c332e2dfe46dddde626faed0bdac5535219d2b4f591829bda58532a52690a53" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/reuse-requested/output.txt", "bytes": 1318, "sha256": "a61512da2c5f62c72fdd904bab9973920c68bfad58d61a352b52175830b434bd" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/repeat-2/reuse-requested/receipt.json", "bytes": 6620, "sha256": "a08d7d7afd82e6af1ae39301ebfa1c3d21df0b74d154d044c1777073de75cdda" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/server.log", "bytes": 55534, "sha256": "52f848bf26e42ea5009d891a00d32473724869839b8e6e90cc3d23d29d16ad25" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/cache96-t14-u4096-control/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.csv", "bytes": 1135, "sha256": "c6f9d46ad5a9b3255978b09ce6998a9207ce63428c102abd6d6257e159cf1262" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/requests.json", "bytes": 68787, "sha256": "dbb5d98c99ec1e87ec0149a25c4cb5b02da33a7b392eda5c27a0b7bdcd9e3863" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/resources.jsonl", "bytes": 329217, "sha256": "377c2ac7b22074c43789ce0624e543e695046f1f899f201d283bbc7149b85f35" }, { "path": "iq3-cache96-t14-u4096-accept-r1-20261005/summary.json", "bytes": 7392, "sha256": "922f2a15c82bd1e27fe1ff871f1cb2a201c907b607ca767af279fcfd2cbfebe9" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/quality-235929/payload.json", "bytes": 2050818, "sha256": "8a53e56f2ed23b00dfb1e6ac4046f9e652c55181d5d2a0156684a124eab8a32b" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/quality-8192/output.txt", "bytes": 699, "sha256": "196a26330140ba8a28d125741071d72c65b562c0480eaa1d00f5350da3dfd78c" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/quality-8192/receipt.json", "bytes": 5913, "sha256": "0c4248148e7dfb6e7d3f9dbbb8a4a6673c410446f824748476a1636deb66e352" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/requests.csv", "bytes": 459, "sha256": "ca2dccc1a0df7e8df6b107e58cd781334ee422b075cd28c8c70064a3eeeb6840" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/requests.json", "bytes": 14071, "sha256": "4c0b93e6ac36788f660eaf6e72ec5fbdc2ad3baa70ac79e3b1d8fc0f06eb8b06" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/resources.jsonl", "bytes": 224343, "sha256": "4e9387bc27604007028aa9e03b5e814ddf68be3600b0486d28afb0a7f712092e" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/server.log", "bytes": 364439, "sha256": "98722632095b5eeaec8f84e684c307ae020e6474942994fba03cae8b8170a60b" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/speed-8192-0/output.txt", "bytes": 1274, "sha256": "3fa4ba5d86d54148a6b1f9108f672bf8f69bd523021eb34c54f012c51eb79dac" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/speed-8192-0/receipt.json", "bytes": 7227, "sha256": "df8f240904b5d26af1c1a116d6d1c7f1d489c92f72423ef7789e628bc83a8a0c" }, { "path": "opt181/results/opt256-cache4g-ple-overlap-workspace/summary.json", "bytes": 7148, "sha256": "d81e16005e2f64a28c0d74ed8db3ed02166a0840fd0895471a30b9ea7f3b9d4c" }, { "path": "placement-speed175-20261005/placement-control-mmap/cache.json", "bytes": 478, "sha256": "c0ca804e71544276c3284af1c0d4219055b8ae332905fb716447beaa9dafff86" }, { "path": "placement-speed175-20261005/placement-control-mmap/configuration.json", "bytes": 1650, "sha256": "fb9c4186871f22d7bce37aad37f53dcfa9fbf0c4bda8f66d124bee4c38e959db" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-16384/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "79c0ab642e2955b249a2ef905da0ff19afd444490eee9a45757b3327e1e4c525" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-16384/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "7084a3d25acc547d7a7dd981efb928bd67339c9cc705786e3e96ccebf82a269c" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-16384/natural-quality/output.txt", "bytes": 639, "sha256": "632148770e173903ae34adc1439be6d1bccc4369b2d724d623c68680cfbac1b7" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-16384/natural-quality/receipt.json", "bytes": 9256, "sha256": "7d423f8d61bf7053c41ebd65d93412483398a415e7a1763da4e609c0d8480da1" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-28672/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "c9edacbacc6afb9c2368c6c0eda0c8b3ae3a0d559e7a521e4b2af5ac3eafd09c" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "7fcf159447f99a72417a300947eae8f93658ccd3005652f90fb0c1e720814bf3" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-28672/natural-quality/output.txt", "bytes": 803, "sha256": "5a0a9b2a0095da77ce12223f37c3f5bab27755bfaa6393d6a8c8ed44a71465f7" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-28672/natural-quality/receipt.json", "bytes": 13499, "sha256": "e0d0f1a0502eedfc94ba4e940ea6b01035f1be2eda6afe7a73b6a8d420b85866" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-8192/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "dec5b093401a35a7fb68a6804762d1169d793817bf5f5af8b1b5db008dce4404" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-8192/natural-quality/metrics-before.txt", "bytes": 2581, "sha256": "5d06929a4798290d124f648c6ff669b465052aefaffdc5eb7166c471faebc930" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-8192/natural-quality/output.txt", "bytes": 721, "sha256": "025800792ce2322d35b6301fea4ecbf635cc1ee1307bfc51f6ef9d2cceb4878b" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "placement-speed175-20261005/placement-control-mmap/quality-8192/natural-quality/receipt.json", "bytes": 7325, "sha256": "a66d2d54e987eaf3d2bce62276a1f6f70c6c5d6e8cc069756ceb8fa691c8de38" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "1c30de5a5afb71dda14d450c248f26f812a353d038db1add63a2f2e2187c1066" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/fresh-prompt/receipt.json", "bytes": 8621, "sha256": "c4b7d9503faeb90d6d68de9498b26fafbd6e887956edfa8b76ae5ae1f1011dda" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "07f12b7414c5803585808b941f444b83f2423e3211d157e97e4272df7815ac4a" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "5580351fbdd7c8da14f0875cb9a291fcb4fe79d9571edf02391988c377c6e646" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-0/reuse-requested/receipt.json", "bytes": 6683, "sha256": "75f9180500a2cf9d181b270f85d25287ad0662378b534c449f123f977a1fa6b5" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2592, "sha256": "c79a6e8de51be8d71bbdc969f581ddd5450622cbeea974b00d2f50ba9f3fa665" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "74f7bc7a661ffab22fa2400c5862d82deed9b9e90196b306fe5d8b4cc6fba1ab" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/fresh-prompt/receipt.json", "bytes": 8618, "sha256": "074d545cc29f27cda87317bfae0f9b21e9e4c59a738e5833affb9f10016c5792" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "916fc17a2210d08d27205d4c0f4fc683794ff231fd2636f8b9a9b4aa7b55bb99" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "52ae1a71ee6b65af6d6a2bda62cfc9a9b7d89c97f9fedbeb8ff4e880c40c5d92" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-1/reuse-requested/receipt.json", "bytes": 6688, "sha256": "9f4d13fa1a5b7e3c7323654e1b55a3932b00a865d869169f709139205f9d75d8" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "ef3af82d170d7e2ca182ec0f36e48b802f99c7b49b49dda93986c0084bc9e04b" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "9e5a4be4b5602b245305c4a61d0f8ff8b2538e86b638e62c1de7c1d9f3465c6e" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/fresh-prompt/receipt.json", "bytes": 8620, "sha256": "ef801d58ddc2e8e2ff70bafca1d26bb5ce7024db2f048d4cda28b993efc6dae1" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2593, "sha256": "f05cfe36ddd114cc2178ae2d5e85cbfbba7d7f8aad72bae550fe79c2a8d3c98d" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "c8e148f2fd50a72f4bfd6e6f58f8f3d7e894149e889e0c129aac07365bb3aaca" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "placement-speed175-20261005/placement-control-mmap/repeat-2/reuse-requested/receipt.json", "bytes": 6685, "sha256": "872093883d7130743eb8d15e0784f89ad47daf419f79ed09e6cd6e233b77650a" }, { "path": "placement-speed175-20261005/placement-control-mmap/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "placement-speed175-20261005/placement-control-mmap/server.log", "bytes": 58239, "sha256": "af0f41d2ce817e5ccd3a5f5c54d63777d1b86cb5622c050b15f9905da1a21359" }, { "path": "placement-speed175-20261005/placement-control-mmap/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "placement-speed175-20261005/placement-none-lazy/configuration.json", "bytes": 1650, "sha256": "a143594576931cc78494a307d838d0347f5521095448220c455e98ad540f42ce" }, { "path": "placement-speed175-20261005/placement-none-lazy/server.log", "bytes": 15240, "sha256": "4a1bbc7acaa4e4b76b07d59f6a7b84598bc0efeb50e5ffe0428f810ce19a957c" }, { "path": "placement-speed175-20261005/requests.csv", "bytes": 1253, "sha256": "25918cc6267fab8c95723ee6f0de892dc5abe55edcf6bb85af81d0a5845c6f5b" }, { "path": "placement-speed175-20261005/requests.json", "bytes": 80417, "sha256": "0ef087eb0abcf80b92316418127253184586b94f9cb17ab67c844298c9926282" }, { "path": "placement-speed175-20261005/resources.jsonl", "bytes": 362752, "sha256": "2dfc3f57f4811fb2dae587ec27226eb3064175b45a715afc0011cf140c130d18" }, { "path": "placement-speed175-20261005/summary.json", "bytes": 9845, "sha256": "97b2af536849938c2ea85ebc0736e30b31a85c104e39ebdca4ec2c9ccfc68e71" }, { "path": "placement-speed175-r1-20261005/placement-none-lazy/configuration.json", "bytes": 1650, "sha256": "a143594576931cc78494a307d838d0347f5521095448220c455e98ad540f42ce" }, { "path": "placement-speed175-r1-20261005/placement-none-lazy/server.log", "bytes": 15240, "sha256": "55a15f4e4de1b08a98d90a48c073a9ac534df7fa389cdbf4ffe71aafb065e044" }, { "path": "placement-speed175-r1-20261005/requests.csv", "bytes": 123, "sha256": "5112c75f4ee7c93c2640ee67e33645dd95d9bcc35cbec384195486aff02824b1" }, { "path": "placement-speed175-r1-20261005/requests.json", "bytes": 3, "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570" }, { "path": "placement-speed175-r1-20261005/resources.jsonl", "bytes": 57328, "sha256": "209a7eb200a1d5d8ef1a520c2efc43f1532ee8a42aa8c999fb77a869b87750e0" }, { "path": "placement-speed175-r1-20261005/summary.json", "bytes": 8148, "sha256": "96d565892876f02a3ead7c23ec4009573372d2ca6119a2ce24e386a655cc96fd" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/cache.json", "bytes": 478, "sha256": "c0ca804e71544276c3284af1c0d4219055b8ae332905fb716447beaa9dafff86" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/configuration.json", "bytes": 1650, "sha256": "a143594576931cc78494a307d838d0347f5521095448220c455e98ad540f42ce" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-16384/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "a3482f5e2ac12183432897899b8b05ed918f621eaae080b0767abb193aaca32a" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-16384/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "da8dc964a9bcff9c6279675de09dee8461efd4a4c2c1c16a2b72e5dff004d664" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-16384/natural-quality/output.txt", "bytes": 639, "sha256": "632148770e173903ae34adc1439be6d1bccc4369b2d724d623c68680cfbac1b7" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-16384/natural-quality/receipt.json", "bytes": 8872, "sha256": "60b77f0d366575333d6f53c49a678df9c206e3994609c252a56aca460ba9ed46" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-28672/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "a95b7511ec2f1f5282137ed22b17a0e302ca80b1ef3ecc24a900d78c7a602fa5" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "bdd2e0fa35c07383a6af649b2bfe54df70257d3d0c6b5bca96ec34a116c44a4d" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-28672/natural-quality/output.txt", "bytes": 803, "sha256": "5a0a9b2a0095da77ce12223f37c3f5bab27755bfaa6393d6a8c8ed44a71465f7" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-28672/natural-quality/receipt.json", "bytes": 12729, "sha256": "c491f0b803d8b52df78ea6d20a5115dc5cfe3d559106d7f27b49bab53997e0fe" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-8192/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "d46a12ea46cc5dad5d563466b7a16e8782daca039c98d1952fcae54378adec01" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "22a638e48f356cfc8eb4aeb273d84e1a2efb6310ef7f7dad5facf5be549974ab" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-8192/natural-quality/output.txt", "bytes": 721, "sha256": "025800792ce2322d35b6301fea4ecbf635cc1ee1307bfc51f6ef9d2cceb4878b" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/quality-8192/natural-quality/receipt.json", "bytes": 6922, "sha256": "1534275418ebf9f24299e8f1c55ffcfef4673e8cc2f6d1f74502369e440bb841" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2586, "sha256": "e1cc3e15955c291c35b28d5ff4f427341af93345ab95f72bbbb71902c8332c81" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/fresh-prompt/receipt.json", "bytes": 9001, "sha256": "acaaf73954c909f32511f0baf766fc1a4e173ebb425fbb2666a9a7522bf439a5" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "46fb6329d4faaf2140aabf207c6e405e2cc94f194e96b33d6c95e23e2508ea98" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "4b957c5ed09ad7d12586b20ae872349ffd978c50752f64a5b80391649200f6a5" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-0/reuse-requested/receipt.json", "bytes": 6685, "sha256": "bd123a00291e58f95fa26221f8973eb70a5582fb08bfa0a6088303fdbb0e1b15" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "7113a07a176a38a5ebd90b24f249bcdc5634c7249944ef3e4c2a74bde41afc83" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "2a7ce9a1237c6224ae62f1ee8a0e3dc5b8b40779478739d00d55b89a527c5c5f" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/fresh-prompt/receipt.json", "bytes": 8612, "sha256": "d603ccfa0546c2e30ca37ed369beba0942185b82ba5e10a266dae27681678f4f" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "1eb6d6c07d551f699b9d87fa1728835a0f56d438858ea341db15cb7ee9378282" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "44c7365411cc8b9fd8b80da2a1cf63830018840f3d1b83798a00666efae88447" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-1/reuse-requested/receipt.json", "bytes": 6681, "sha256": "87769f6ce6c4abfab2aa72b2bd54f04e7c95818d8a34e3b9860cb1b9c1f7820d" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "9d3563d3f4121146f3ca5e586fc246b0fcc3701cae70a63e2b649fee02c8a4d9" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "41e071c68ff44f6413442c1e4c6a7c75f6112aa80c901c47cd3d1c3236696bea" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/fresh-prompt/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/fresh-prompt/receipt.json", "bytes": 8619, "sha256": "525c5d8e53986ee545eed499f8d8b5c846252589488d78888267a64f0f1a0215" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "251c02d8f02b0d5e97b3b49b8ea7496a7f4c92a39d95f74befd495318f58e318" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "8464bd6d7abca2252743bf08926f4e75ea3e9ab8cfecad2cb75c998c0542973c" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/reuse-requested/output.txt", "bytes": 1380, "sha256": "fe0e3a74fc7119f7789b1051e5268a1263f7708cfb01815b610572a52a552113" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/repeat-2/reuse-requested/receipt.json", "bytes": 6685, "sha256": "5c1dc3831d3da0939feb44611d6763d01d07c0fddc7fecb47b78ffbc58fed1d6" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/server.log", "bytes": 58022, "sha256": "333138a34ff1b46e62939167fc858e2f3a053b81fbb6cad22377fb0da3fcd9b4" }, { "path": "placement-speed175-r2-20261005/placement-none-lazy/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "placement-speed175-r2-20261005/requests.csv", "bytes": 1226, "sha256": "e4cd0a061afa26d82ba2f4d4f94d5ea4b605bf1e8b26462829f985ea6bbd738c" }, { "path": "placement-speed175-r2-20261005/requests.json", "bytes": 79138, "sha256": "8a77531a7ac32a671d7eafb3e8fa5a1b98c6d135587750d201eb1bf2ea2ae4f8" }, { "path": "placement-speed175-r2-20261005/resources.jsonl", "bytes": 404282, "sha256": "904a6d23eab6aca3ba543a6d2d267894602b255320bb7520e4063139a5799450" }, { "path": "placement-speed175-r2-20261005/summary.json", "bytes": 8078, "sha256": "e6fcfca339f3f477f957f55822dceec941db3206468c6ebf25f0ed69308fb74e" }, { "path": "prefill-dma175-20261005/requests.csv", "bytes": 1452, "sha256": "7cadc96ca164f212577c6d7c28f3f27c7e89a2761094efc0c3f5d906330e388f" }, { "path": "prefill-dma175-20261005/requests.json", "bytes": 71966, "sha256": "2f572732c568b0b27408a2831e282798c0b2bf8c62f5e4f9e1c24fbfe692c049" }, { "path": "prefill-dma175-20261005/resources.jsonl", "bytes": 320236, "sha256": "ba68f3ed2d17ea02c28b70f6b56c71e6ed7ed9ef25b79b60487b25388e468d27" }, { "path": "prefill-dma175-20261005/summary.json", "bytes": 11354, "sha256": "b6b9ba5307cc1016156bfb4cc0767c9d1fdcb36af3c980583b1fc77fc7de6d81" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/cache.json", "bytes": 755, "sha256": "70a3f293a511ab68727450d59c9fa78c84b0e95b2d7f3c291ca6faef43300a91" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/configuration.json", "bytes": 2173, "sha256": "ac4d4df09305f0d1ea50f9cdfd7559869fd93c5ccc0d9c450d82fabec915b75a" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-16384/natural-quality/metrics-after.txt", "bytes": 2917, "sha256": "1ac7389bd6aa24adc04bfe8d38c150fcaaa001d5afaabeeac055e488c98903be" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-16384/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "97454e3a092078aac5ac64410399fe0626b125b138f94c73a7a4d511ed1dd4d8" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-16384/natural-quality/output.txt", "bytes": 598, "sha256": "e8a69a90a373c437e1376e2712c867e8240e98106081f74960f4b8ebe980ef2f" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-16384/natural-quality/receipt.json", "bytes": 7932, "sha256": "10ab4166149926ca2ed761fd79847d577dfdb6e2ce2ec31649c820e44a39b1a1" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-28672/natural-quality/metrics-after.txt", "bytes": 2918, "sha256": "1cbf786a2fa3b7e3cbf6525066cae56f289edf3368fc07fe2cd7ce04b2449b77" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-28672/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "33f1cde65d731926b6748daf74ca6bc22622e30d3a274e0ea157d5295acf3464" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-28672/natural-quality/output.txt", "bytes": 600, "sha256": "9ecf826b60fc5c9d3c41281a033d2ee44b5a30b021e4db69a8e673108880636f" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-28672/natural-quality/receipt.json", "bytes": 10623, "sha256": "f60e3d6c0d2bd6e6fea728e093c6c5cdb7cdca91d1a185c982f93b7f018e3f77" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-8192/natural-quality/metrics-after.txt", "bytes": 2915, "sha256": "6dfe58d7b0c4e5914e5b68ff613bf4d7a9dbafa5ace60f66858e401a70c921fd" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-8192/natural-quality/metrics-before.txt", "bytes": 2904, "sha256": "c652698bf79bffb271ad38b5e1bb9b44c915bd91697d61ca1cf6b91d1386bd94" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-8192/natural-quality/output.txt", "bytes": 559, "sha256": "bcc9b36fb5343f2e93238c2a5ed3b9cb844eff09ca703a8e7276bd417f3dd654" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/quality-8192/natural-quality/receipt.json", "bytes": 6778, "sha256": "3f813c26c2edf6b530329ad9397d25325902fb2e73c84c7d16b19de32cc0937b" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2906, "sha256": "d40c84e15d588ffb82d9711e0195aaa6736d8df00dc1043d7ace6014c2ec8d93" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/fresh-prompt/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/fresh-prompt/receipt.json", "bytes": 8073, "sha256": "c4dee0cda3347ea90eb2374952468e306cdc89b2dff7eb3a37064deb3f0fcda3" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2910, "sha256": "f4d4b26bde1aec38b98735cd0f905122236d21fc7d3960179822256415d25eb3" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2894, "sha256": "2222fdf5a8e2c7f9047b4bbca9b1b5ef254071acb482a0242fba17e6fa057b5d" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/reuse-requested/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-0/reuse-requested/receipt.json", "bytes": 6528, "sha256": "e5f16fcb52b922bad9a1fb2c82cd0e9cc6d8d800226171b5d4a4eb31246f98bd" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2911, "sha256": "dc1c18bae5ea3aa9e6330a2eff9cb250e73cdf0987051811794375b35b58b83a" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2898, "sha256": "dfbbdede8a5db520182144ca2eba9bb2c48db702cdfee7062ce67ce0ef56a535" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/fresh-prompt/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/fresh-prompt/receipt.json", "bytes": 7688, "sha256": "cb5a1e24132785e73c675f7caa152a5fc640e8f3b8753892cdb589a739061e07" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2914, "sha256": "3c5d8651a0f9e0efbd93671e4981925fcdcf07afae792ab232bf697b1603447f" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2899, "sha256": "e7d2ddc4546e44c1577d35d1481d75e09c8fdeac477e44afcebd6fe100e32649" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/reuse-requested/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-1/reuse-requested/receipt.json", "bytes": 6524, "sha256": "fb1faa002a7ee5f2b639e77addbb7ce1ec99976d275bfeb5bf9bc6343741790a" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2914, "sha256": "a64aa825c78e8f81fe362bed6535db4ea7ff30bc4247592ddc18d7175330585d" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2902, "sha256": "9fd0797c1e188e0a51b3f26f668e41cca1dd96c85f8ea60b6f36156ac0a1824f" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/fresh-prompt/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/fresh-prompt/receipt.json", "bytes": 7691, "sha256": "0a99aabc2338c480f003ca6fd99969d504b4515b12870661463b264f13c4d9f6" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2915, "sha256": "9c3b3bb6e4f78abeecf27852e5f64da04e95a69c38a8ae85d587705afd682b53" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2902, "sha256": "258048048d1bf9431085085fd76f4f0f26affac1498c703a265900f17d37ef4f" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/reuse-requested/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/repeat-2/reuse-requested/receipt.json", "bytes": 6529, "sha256": "a793fc03e2c4ea6efdc38113b77edbcdfed7fbedb494986d77ee44e9fa6cfe53" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/server.log", "bytes": 75697, "sha256": "52d334a24af2bfba51aaa36346513089f8bd7e2ffc2fe3cfe5a52449c382c70e" }, { "path": "prefill-dma175-20261005/uncached-hybrid42-mtp2-pinned/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/cache.json", "bytes": 755, "sha256": "52307ba560c9174e61e7b02609095517206e236a43e1356c54d7c1482cce13d5" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/configuration.json", "bytes": 2171, "sha256": "efcf44f8d07f15a7be2b26bf68ad5d3bf578976ac2138fa6887b72c012f4aded" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-16384/natural-quality/metrics-after.txt", "bytes": 2919, "sha256": "b36a103944b992bec71dacdcd8f5c4287c135db9d00d236d313bf6bf373b45bb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-16384/natural-quality/metrics-before.txt", "bytes": 2906, "sha256": "8cab50f063a16e719bfd59d16bbbf2af41cf4c14b6dac2bc360cb837ad4ee922" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-16384/natural-quality/output.txt", "bytes": 583, "sha256": "6acd1397896d3ce53d64b8174c5444f96f194e2a49682b6e39dae50eebf601a8" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-16384/natural-quality/receipt.json", "bytes": 8018, "sha256": "4f36aa5d82ceac2c5b12fda9a8a764fc8da35983927a7c573ee9d3a21573efee" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-28672/natural-quality/metrics-after.txt", "bytes": 2918, "sha256": "f543355e58a3daa47660f02507430ef0622f05756de10e735b62fd5fe9c70ec6" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-28672/natural-quality/metrics-before.txt", "bytes": 2907, "sha256": "324ee5655333665691291dafe85aa6345f92ec710477a1edd0830bd7c460e18d" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-28672/natural-quality/output.txt", "bytes": 584, "sha256": "77b697fff946a19c4623904d709047f95f576357b171107440c629b0471af687" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-28672/natural-quality/receipt.json", "bytes": 10694, "sha256": "7406c57a9553fca7aac713ea4bc032ca508505da23f6e099ff2e91665fc25e4a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-8192/natural-quality/metrics-after.txt", "bytes": 2917, "sha256": "dd8faad5fa429b71aef83007ad8463b69787611859ddfe6079d82b7ea0cf110a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-8192/natural-quality/metrics-before.txt", "bytes": 2906, "sha256": "30bfe3f6ba84659bc62f59eace92bf86374d3e06dc4f2e1e117aca185e3f198b" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-8192/natural-quality/output.txt", "bytes": 746, "sha256": "7100eef89795c736cd23cfffc8f77e6bb36d6f85445794a2fa5de8d676c73b8e" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/quality-8192/natural-quality/receipt.json", "bytes": 6849, "sha256": "414dc5d948f67046ee42df416f60aec9877f545d09ae1ac3767ad0efe92a33bd" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2905, "sha256": "bdb5b6c46e9ff20008d73b0f8793b12899481c3c15786572d50e7afd15e74305" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/fresh-prompt/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/fresh-prompt/receipt.json", "bytes": 8530, "sha256": "4c43232a3603563bbec2e7a5ffa1ab0df6eec207a9e60349f2a7243170e27546" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2909, "sha256": "78fe127a0dc2e42b824bfd6cb1ee962b7341b0e2c39d6029afdab05f52f7d4ab" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2893, "sha256": "3e7770e675ed685e804e6e04e7ef4b363b5d502b2d547c9a9dd2f2a20d91df56" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/reuse-requested/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-0/reuse-requested/receipt.json", "bytes": 6987, "sha256": "8ef8115153d2e48133e9d4141b010d79e5394858165e950264ed10aeae73f541" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2912, "sha256": "f0f1998aba50f820b74234754171b3fc47433d02cb59d2a534c2461882d4394a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2897, "sha256": "d36c2784e4f0bf592e7f5e0b74cb57d10c571209712547e1a55e539b15e18628" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/fresh-prompt/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/fresh-prompt/receipt.json", "bytes": 8152, "sha256": "54c592f3c61d473eef0454061b432e41cbf21944a3f2191389fb1fca1a19016a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2914, "sha256": "4d65b13ddee3a38b6ec0baba92348f69e825f4b746691cabda44e98786ddae5a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2900, "sha256": "5880ec682d0963e6fdde32822101fc83151cee88604eeadd96fcdc079afe713f" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/reuse-requested/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-1/reuse-requested/receipt.json", "bytes": 6988, "sha256": "5af088ddad6f482581264cd1f4c72f9e4637c0ae534a2dd2a8db1b0449fdf2d0" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2916, "sha256": "ee9cb211c198ebef18050131ce055203c6be191e5a7a35cc22568335c895babc" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2902, "sha256": "3d9c4b363226bbcf8c168f4a582f26f72336f57d10a722ebdb4431eddb925efd" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/fresh-prompt/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/fresh-prompt/receipt.json", "bytes": 8146, "sha256": "40007a1bb1ab15769b63be69eff8481acb3b03333f8db4af99499c04293ebbf2" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2914, "sha256": "67c509677a4d8da6fc6220646840772810774841e6a5d180894b7f33321b64b6" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2904, "sha256": "c0b69ebf9b52bab0057cd2fac0ff573ec73067dab850c1910b367b127a2fd5f0" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/reuse-requested/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-2/reuse-requested/receipt.json", "bytes": 6990, "sha256": "9c7c1414cb36dfb608d924706a51bc31e2b83f669a2cb7039e39fd33a6354eca" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/fresh-prompt/metrics-after.txt", "bytes": 2916, "sha256": "e9969f416ec1f0f6a14352a3eb67a4f333d951531d149f3e39b0ade84230605e" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/fresh-prompt/metrics-before.txt", "bytes": 2904, "sha256": "17a56b5d1936a73efa510650e078cd397d72736c29d900ec7e16d332360d869e" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/fresh-prompt/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/fresh-prompt/receipt.json", "bytes": 8149, "sha256": "e64397965d2b75457dde1b1f20fbe5d8d4406018080398e79179f912a94cdbc9" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/reuse-requested/metrics-after.txt", "bytes": 2918, "sha256": "e82219e068bdda943b39b417bf13ae1e3e4ba490a05472c1bb41079ceede98f2" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/reuse-requested/metrics-before.txt", "bytes": 2905, "sha256": "bdca1673431bf3ebd283c8c631ad2308ac9fc65fd4c576b48cbb5cd8c7a805a7" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/reuse-requested/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-3/reuse-requested/receipt.json", "bytes": 6988, "sha256": "a416dd299508d988accbaf40f59f68df3a65bfcd6faec9712ae549e0f3ef9a87" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/fresh-prompt/metrics-after.txt", "bytes": 2918, "sha256": "1a92d124ce4e0a7796a78fb3add20382e34794044dd42b4a9d815f2ff0ac8a5a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/fresh-prompt/metrics-before.txt", "bytes": 2906, "sha256": "ccb7653caa9750a7c009c48bb2224df7524ebace61bfd5e66d9d7d4eb7193d4d" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/fresh-prompt/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/fresh-prompt/receipt.json", "bytes": 8144, "sha256": "5cfb2bc70e3dc8a924a9258ac4c68d7b499902b81fcf2a4c0e6201aba4f43299" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/reuse-requested/metrics-after.txt", "bytes": 2918, "sha256": "2409feec8e8a8518ecc83b37a8aaa2dcd0afac27e2df04dcfaa4a8b90e9b7002" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/reuse-requested/metrics-before.txt", "bytes": 2906, "sha256": "ce57b33aeb5e910f0bc782f2e3546a029a4f00c26cac38b98ef66a5a69d54c15" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/reuse-requested/output.txt", "bytes": 1353, "sha256": "89a4b56e4b7f938d49b8975e19439fae102e88d8bb8914ab3a54bcc3f27f6982" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/repeat-4/reuse-requested/receipt.json", "bytes": 6986, "sha256": "c2af9131fba1377599aa0adf7e7bf3404ca4ce83276a0499d4753d8d31a7b5ef" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/server.log", "bytes": 91205, "sha256": "553490b93fed20a74bc052da61bde409a76ee3007a9e4157c813a3e7bd9cb015" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-f16/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/cache.json", "bytes": 755, "sha256": "25f3fbd35a7caf8eadbc6fd80b188f04e31c9b2842feef4de2d94c48586a1281" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/configuration.json", "bytes": 2173, "sha256": "12424a7e77d263f241d71201b3b444bc4a48910c534f9a56b66db1836b1941b2" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-16384/natural-quality/metrics-after.txt", "bytes": 2919, "sha256": "bc8803114831b00c5777956704e385c578a0bdfecd8d160c7d3df53f3b788585" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-16384/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "e1eb8a30324754786923e44e342e4c97a32180bb569e9f931f9660440641d7dc" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-16384/natural-quality/output.txt", "bytes": 590, "sha256": "eaaded4ad776a5775cfe44335091e024bca64dc6391c1f132f8c74c5483cba03" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-16384/natural-quality/receipt.json", "bytes": 8020, "sha256": "f92d9b87535d668102de5764c9ca81675bb60fe257717afd64d1b804157afe57" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-28672/natural-quality/metrics-after.txt", "bytes": 2918, "sha256": "0b832f83e0be7c56899f9e084f0cc455a4bc98a73ce710d9e8b93ed8a97d289d" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-28672/natural-quality/metrics-before.txt", "bytes": 2907, "sha256": "21ea84867cb90825f7ca397e75e755239fc7af975ae40c327c8cf52f466d19a9" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-28672/natural-quality/output.txt", "bytes": 734, "sha256": "1abca96c80b94cc830126784008159114d1fdb8d77303060987f85eff8598b90" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-28672/natural-quality/receipt.json", "bytes": 10720, "sha256": "b0c304083c9374689489924cd9f48f293f4e557a36b55ccd861e50256b5b8da4" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-8192/natural-quality/metrics-after.txt", "bytes": 2917, "sha256": "8c808df191e2084d2dadf16b23009e890d6367d850aa10607fbd4b6b10a8f038" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-8192/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "7415d3382f19873d9fc48500a201b0f32cefe05f3da0a1527e1593656f678831" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-8192/natural-quality/output.txt", "bytes": 559, "sha256": "bcc9b36fb5343f2e93238c2a5ed3b9cb844eff09ca703a8e7276bd417f3dd654" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/quality-8192/natural-quality/receipt.json", "bytes": 6853, "sha256": "1c1e173d57114c0c1600e61dd326082a86bd177eb352b72498703c7e5f5cfbf7" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2906, "sha256": "6aad1507a8574abab796c05150af93bcdc3ba9671170f790f75fbf54b432ac4f" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/fresh-prompt/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/fresh-prompt/receipt.json", "bytes": 8535, "sha256": "0241a0a269584c6fba807b7675541249f975f3c2f1e1d909a0dac726ac5ce4eb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2910, "sha256": "525e79f07217e904625800fa52ad27e8d70ffb177ad6cc0e7d84678b258af453" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2894, "sha256": "2697049d897afd80140a647fd95c07677cbedd6f0021f8aa86108c286f36d531" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/reuse-requested/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-0/reuse-requested/receipt.json", "bytes": 6985, "sha256": "f4ff30666ead55138d8f0c920652be4bc9fd265c5943a623b336458b0cfd89fe" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2912, "sha256": "98bb73e6c9e212ee141493a1a39fbfd2f5e93c2871d5d191a59b90cc7234a316" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2898, "sha256": "d0125be4225c863605a504f47652d4bbc9cbf8e7c740317047f8cbea4fb06e08" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/fresh-prompt/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/fresh-prompt/receipt.json", "bytes": 8533, "sha256": "8a5bfb14bb82ac4b63c8effd921cf06382fe42af704ac5d8dc9c4b1438d8f363" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2913, "sha256": "a34b78327a8dce0db194918a620a5d61be51af1e958e718eade3ee16449f22a5" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2900, "sha256": "4c5cb9ebaeae7f4df10e944dc0749311db3a723b01fc60eeb632dcfff46a6fe4" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/reuse-requested/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-1/reuse-requested/receipt.json", "bytes": 6989, "sha256": "5358309c93f25335bbab7406d240a2c07c30b772d2a677585fcd80c053686a76" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2916, "sha256": "67909f0b8d1a443abf694e1b82be8865cdb14b65c357d7efe41d2a2688c84a7a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2902, "sha256": "7b9e2003de6b5f9235d3b965f5fb367833a4da94cbc3747ee966d281d3548819" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/fresh-prompt/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/fresh-prompt/receipt.json", "bytes": 8530, "sha256": "61e7723f1641deef345b9095d3c05bdfc6b206ca3f3aa5cb9bac212679923d83" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2916, "sha256": "90987e03d6b60a666ecae6e40e65e8e1514d34bc9c59a19b9fc6d1b6a6b8895d" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2904, "sha256": "a6f9c868696e1f71b367d17edb894a99dc04f463e7dbf8999a33c8194ec2aa18" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/reuse-requested/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-2/reuse-requested/receipt.json", "bytes": 6985, "sha256": "0c15aacd3d0fc04ceabc0cd55e36e2175c957ee89776f5c5ff084bca6256d611" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/fresh-prompt/metrics-after.txt", "bytes": 2917, "sha256": "c5b59355bb6c2ac9759a378963cd7137f864411967c4777a3f9b1c8a8343c4db" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/fresh-prompt/metrics-before.txt", "bytes": 2904, "sha256": "508d6a3eec70a8d9e5b80c810a20f8c706078c6e991a0a40dbfbfa6812d4ac0b" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/fresh-prompt/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/fresh-prompt/receipt.json", "bytes": 8537, "sha256": "b71fe1cfd8ef547ac4b5532fc6b51200b05b4afbdd6157ae0564d6c11c0d6aa3" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/reuse-requested/metrics-after.txt", "bytes": 2917, "sha256": "0bec70647a49e33d82cc9150136612d1f05e0f67a8136fde04eee2b49e79a1ae" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/reuse-requested/metrics-before.txt", "bytes": 2905, "sha256": "41f4deef5ebc14ebc1a9348b3bd3f22be60c937932cb419f9e13325908984cc9" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/reuse-requested/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-3/reuse-requested/receipt.json", "bytes": 6987, "sha256": "112f06c0889a3a3ee1d449c4a531d2286ebf79782a3b8916fbc5b18e921c94e8" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/fresh-prompt/metrics-after.txt", "bytes": 2918, "sha256": "8d87336102bd76beacecc11f888b1db8716bf9d59c71cc11c487f09d6de62edf" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/fresh-prompt/metrics-before.txt", "bytes": 2905, "sha256": "6f2c4899dfda8fd3c405a2be11f604cb2b79657c1e88c55eb42a9fe2985bc685" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/fresh-prompt/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/fresh-prompt/receipt.json", "bytes": 8531, "sha256": "f40cf026399c09ceeab2577ed6b764b761e6549f74699bb8ac160ceb2c5b7108" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/reuse-requested/metrics-after.txt", "bytes": 2917, "sha256": "b55275ac2e5c6c81e1300b6a823a5337d71ba7ec6455fd62f8723ff6099fd8ea" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/reuse-requested/metrics-before.txt", "bytes": 2906, "sha256": "fd748251765d1497afc706a25a4e9646c2ff436f95ef37f9e33e2df3235360cc" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/reuse-requested/output.txt", "bytes": 1500, "sha256": "cfb5f2aab528d2c3170337cfcbab21cf9f22e6985166497562125c2f2563523a" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/repeat-4/reuse-requested/receipt.json", "bytes": 6986, "sha256": "0e1dfc4ae77e1d191675f7cb6ded50cda348105d3e316fe0a42adfd7042d55cc" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/server.log", "bytes": 92261, "sha256": "1e639b383d11a1206261e72dd9190340456eabea84373299de3edd79b3b12037" }, { "path": "prefill-finalist175-20261005/pinned-hybrid43-mtp2-q8/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-finalist175-20261005/requests.csv", "bytes": 3823, "sha256": "c3414d82d3ebf19df1f911fc60bb57145ca67113e0d05768ec3a2b6f3b49ef02" }, { "path": "prefill-finalist175-20261005/requests.json", "bytes": 215409, "sha256": "2d413845848d39073419f33bfadafa9de9e7c748a529e4730daa6e8995803076" }, { "path": "prefill-finalist175-20261005/resources.jsonl", "bytes": 853653, "sha256": "af7a3d00459d16f67c30e03c124a8df7e611fbdda76ae0f0b04ed1e128ac0317" }, { "path": "prefill-finalist175-20261005/summary.json", "bytes": 11422, "sha256": "3ca2fda59b6b96253c509771c38e0406e21859326a542c51640d7f71172cf3af" }, { "path": "prefill-hybrid175-20261005/requests.csv", "bytes": 2480, "sha256": "99cce8d77b62b2bc4230fe465b4cc87e27d89c0a373174b7967c1abef6b13cea" }, { "path": "prefill-hybrid175-20261005/requests.json", "bytes": 147056, "sha256": "6c5806187457628905bb4cc898c9a6922bf369f7a61007c9e48c369947cd4be0" }, { "path": "prefill-hybrid175-20261005/resources.jsonl", "bytes": 568730, "sha256": "5639f0318d0dfe88b6eef1631adf97a2a34fde3749cab7a95d783cf680f98d1e" }, { "path": "prefill-hybrid175-20261005/summary.json", "bytes": 10507, "sha256": "f0fefde7811f4a123a12a13d0b9a4aec6c8a5a30836d4f34547d6cc30a36057c" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/cache.json", "bytes": 837, "sha256": "00bb6947a291604084f3a7a4db39bdb6cd1339a5bbcf63871758759fc8a05ef6" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/configuration.json", "bytes": 1649, "sha256": "c48fbe8bfcb4ad25e7d1d3be3c5917c398cd08658abce552421eec11f4a8bef7" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-16384/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "49562448a0f9839c0f03234f236f910efba6cd0bc00d259fc41332be0851166a" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-16384/natural-quality/metrics-before.txt", "bytes": 2581, "sha256": "22e6e996ef95e0458c819472063c403c6a6370c3af4c3d69e5a67d40900ecc76" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-16384/natural-quality/output.txt", "bytes": 803, "sha256": "a96eda1fdc5a2a87d425867f852bea32e03ce645827656df48fb50fbd21243f5" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-16384/natural-quality/receipt.json", "bytes": 8090, "sha256": "fa2c7e9fe7b9ade6d1b0a3eaa0fd0d412d19ba476a7d03d3bd716d4437b16964" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-28672/natural-quality/metrics-after.txt", "bytes": 2595, "sha256": "e75c9e38c172b0aad932c761e22a854211e5ba0b763ff9108d626a188c5c094a" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-28672/natural-quality/metrics-before.txt", "bytes": 2583, "sha256": "af243264029c9811445765d805015f86792855ad64b4fc2b915d889617a7361f" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-28672/natural-quality/output.txt", "bytes": 803, "sha256": "79ae0a5b486c8d5fddc1f2ccd7a538308655ec37204dcedb031dcfc0e706d753" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-28672/natural-quality/receipt.json", "bytes": 10791, "sha256": "ce738d64924ad3a01f564c6d62ff53faf6c36c5606c4fa04e711ca8cf49a245b" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-8192/natural-quality/metrics-after.txt", "bytes": 2593, "sha256": "88cdef4cfb050a6a3ae70b7260dc285386a36ef71664153939e58240033aba76" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "bc833054c49ba0bfd27445d06061e588a229bf067042980196823f94297e0c5a" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-8192/natural-quality/output.txt", "bytes": 644, "sha256": "286f0daa6f8e0d6d5f137c600dc83c4da85cbf95e8176e83e2c1dcf293333e88" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/quality-8192/natural-quality/receipt.json", "bytes": 6533, "sha256": "2e693cdfbf6f9ae92b4ed6f9bf803c09ffaf28139c4cec74567264e8bb5430e0" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "8594d56dfe41df6b0bfc125d32d56d12271e34f4c3ce9756ff1feb95a67e01d3" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/fresh-prompt/output.txt", "bytes": 1175, "sha256": "c88df3a85fa844799f818c6b90d8d996a4be90eef913e4a6cd396e1aa9638988" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/fresh-prompt/receipt.json", "bytes": 8214, "sha256": "1101e73c83720a6212a01d860e6a10178965877507232fd8aa43e1813fa22a78" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "7d10c90717df33f90d335f852ebdbbc9985dbec998cd0a21a7d32a25fb80dd2e" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "5ddb887b12faf9f929296211de0f9a644e1111c1e5342e197fa4844e6e4d8010" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/reuse-requested/output.txt", "bytes": 1175, "sha256": "c88df3a85fa844799f818c6b90d8d996a4be90eef913e4a6cd396e1aa9638988" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-0/reuse-requested/receipt.json", "bytes": 6671, "sha256": "1d55994b380e0bb7334a6c843fef4737b424d96e5ebb06e5987102d916ecd3f3" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2592, "sha256": "56ef1704afddab663a46c96c8259990d204a00d5b89e7b4c5811e2c93b07d91c" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "0f9e84e4db6cd04fd4dac18cc28337aa16367f35fb975afc0c35acd6dab9c971" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/fresh-prompt/output.txt", "bytes": 1175, "sha256": "c88df3a85fa844799f818c6b90d8d996a4be90eef913e4a6cd396e1aa9638988" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/fresh-prompt/receipt.json", "bytes": 8208, "sha256": "d34b050665d18bedb9d612d4bc28f17a7fe305e71ad41e7a87583da88e0a41b8" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2593, "sha256": "d1c9a69a8f04fdaec04e555fb57eb1a99ae7e40ce794921198de91f29ab5aefb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2580, "sha256": "b650094e6e029b68236f43d63d4a4e3fc9bacc0a45956d240701b5de9a5e2c37" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/reuse-requested/output.txt", "bytes": 1175, "sha256": "c88df3a85fa844799f818c6b90d8d996a4be90eef913e4a6cd396e1aa9638988" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-1/reuse-requested/receipt.json", "bytes": 6669, "sha256": "4dd890203bbda9a4c69c018c694b3d86c338c39757ecd2e87ffb8863fcefe4df" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "6c52ca8a5d6553cc4d837003f4aa958b861ebbd413d6b768516972ed74ebbf7a" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2581, "sha256": "2dc0acea1c4c04d95dbf7edf1f373892cf2fe3cf3ab48ac426c6f442b85138d3" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/fresh-prompt/output.txt", "bytes": 1175, "sha256": "c88df3a85fa844799f818c6b90d8d996a4be90eef913e4a6cd396e1aa9638988" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/fresh-prompt/receipt.json", "bytes": 7827, "sha256": "450fd8baeb06132c32035715038044199ef5b63887f445f04a5f229083717d50" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2593, "sha256": "ef3e92368d9f27f037bdda996ac519afbd76761bffe76ef342c6f307e18cd013" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "8ffc6d8f506f07e274d3c8549f9334dc703b4d2809a1dfa6c99b3ae080cfbb0b" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/reuse-requested/output.txt", "bytes": 1175, "sha256": "c88df3a85fa844799f818c6b90d8d996a4be90eef913e4a6cd396e1aa9638988" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/repeat-2/reuse-requested/receipt.json", "bytes": 6670, "sha256": "a258814ee52a692d164ff90b75aec5cd4c6a593f6bd6f0692825c5f0b734ad8a" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/server.log", "bytes": 58039, "sha256": "bb2c421e73673193c886beb5464fb277cc11cb82b0242854aa6d344fb48f069c" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid38/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/cache.json", "bytes": 837, "sha256": "918f321040268e5479d182137da6036ce7d779cf7c47ffb6d842fe8efd88c4c5" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/configuration.json", "bytes": 2173, "sha256": "9328fdac61cc49145f4c01e2ce8a7027410b353089688b2013c6798d440eee8f" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-16384/natural-quality/metrics-after.txt", "bytes": 2918, "sha256": "5ceb85ce5bd7654857258ed47414e6528ffa688dd8ad025b09a4407e02bf42b4" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-16384/natural-quality/metrics-before.txt", "bytes": 2905, "sha256": "c59e9c442bb7342bd6925ed58ba192eb5ab66441da5270cff6f2ac9b8b2f44ef" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-16384/natural-quality/output.txt", "bytes": 598, "sha256": "e8a69a90a373c437e1376e2712c867e8240e98106081f74960f4b8ebe980ef2f" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-16384/natural-quality/receipt.json", "bytes": 8254, "sha256": "993cc04ac9b7b5eed515c6fe7acbcab250d77a716ea7b402c3f0369d4be1df39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-28672/natural-quality/metrics-after.txt", "bytes": 2919, "sha256": "e406dbc9f2acd467210409d8f88df378be8ac2b49d85b252ae118a0f4a17215d" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-28672/natural-quality/metrics-before.txt", "bytes": 2906, "sha256": "06e4e76fd66a1ef85a6fb7e5140410d47d58a85d95442c45e2a2dc7438ed4646" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-28672/natural-quality/output.txt", "bytes": 600, "sha256": "9ecf826b60fc5c9d3c41281a033d2ee44b5a30b021e4db69a8e673108880636f" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-28672/natural-quality/receipt.json", "bytes": 10939, "sha256": "fbff0b7dcb62131d866d46d70da59c666ba59c61e43f6d62f62b4209fd04fcc6" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-8192/natural-quality/metrics-after.txt", "bytes": 2917, "sha256": "7f849c22bbbc5eeb7ddd584131a042f96a3ddce157f89da5a7d85d74b72a2fb1" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-8192/natural-quality/metrics-before.txt", "bytes": 2901, "sha256": "5b5ff302eddb5a370f68b9dcd1ed40bbeef23a98ef86eee6021b3d0d8f1c9b11" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-8192/natural-quality/output.txt", "bytes": 559, "sha256": "bcc9b36fb5343f2e93238c2a5ed3b9cb844eff09ca703a8e7276bd417f3dd654" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/quality-8192/natural-quality/receipt.json", "bytes": 6717, "sha256": "f0f5e6ef6f1e2598ae2b057e979f599382ea7e3d5c4f52b8779e89e3ee8c1d6a" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2905, "sha256": "0a21ec4a6ebb1b9fffb157a00c387aef4c1d66c5214a1adeebb9f36b72189598" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/fresh-prompt/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/fresh-prompt/receipt.json", "bytes": 8394, "sha256": "15ad6c69312329f30ef8c1f14df3bc9cf4398273b35dd7c84d74ee0bbb6112b6" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2910, "sha256": "13e2872d0aedfe22e89bcc2b35ebb0a8a3b275768041fb6163d34153c4e1f3e0" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2894, "sha256": "e3c2c53b04207c89a311c9f125e551e5300dd0bac677017ae4f4c8ae1853a9ec" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/reuse-requested/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-0/reuse-requested/receipt.json", "bytes": 6467, "sha256": "9cae6d625c3a2df9ff22aa53671f53f9282d92a149568a3151ca482db1e8e8b4" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2912, "sha256": "bfca850b9a95cb3cb9bb7d89be756fd6692429dc0e149348585bc1ebfc4573e5" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2898, "sha256": "b525e00093dba0bd598364fce67150d33b07d348a47abb31b4676dd9a19abfd7" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/fresh-prompt/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/fresh-prompt/receipt.json", "bytes": 8009, "sha256": "e2f7baa01598a0aa2df57cb34c58f16bf61002c5bb1a5e09588c90b452ff7b39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2914, "sha256": "16dc4160d58a026444ca66d7bce91d7f6af2a121cc42549b89214dc43ab736fe" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2900, "sha256": "434af9c5084e0c92f07e0246abda8ae364ebb2f1d4d49ee2018e3dd1325f30c0" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/reuse-requested/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-1/reuse-requested/receipt.json", "bytes": 6467, "sha256": "d3e7c7ac88feae0a18d566e594a7014083cc86a4c2502d49cb8910e61c20a0c2" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2913, "sha256": "8629d2bdd67573ba4a53887cf31a19d346350c0076de0fefef4e9a30b3d604b3" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2902, "sha256": "c8dcdb519e4245246cbdb805e4387a99f89014b04c25190d9b7aa05b40489817" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/fresh-prompt/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/fresh-prompt/receipt.json", "bytes": 8015, "sha256": "4c6ae0009301be5d5b1ffb138f92144192250df22baf37098760420b1a00fb18" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2913, "sha256": "6cdf1c4f28d9832ddcbddc8ab749addb45df78766949999b724e326fa3c94084" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2901, "sha256": "f4980a5c2e155805f8fc59122ca7d806766f9fcdeb0e2f1bc57347f33eac6bd3" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/reuse-requested/output.txt", "bytes": 1044, "sha256": "9158a9c98faf0e2d57dc4c9b4d4cde0447c1c0997276e124e0d26e66b1b80536" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/repeat-2/reuse-requested/receipt.json", "bytes": 6468, "sha256": "73d8669a3734c1311aab71b7d5a10a15ebcbac9e26dcb100e38518206c5ecbcd" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/server.log", "bytes": 75912, "sha256": "9729336e688ab0cdad6753d78df8b40c551aca55d88ec0fba6d2724e8828bd68" }, { "path": "prefill-hybrid175-20261005/uncached-hybrid42-mtp2/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/cache.json", "bytes": 755, "sha256": "37c74f08ef5e8766d75e7d23d03ba37be2bf1b4756e6e44d8ae97f524ee593f9" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/configuration.json", "bytes": 2171, "sha256": "59c0bdd22f64f5c7e277be3c4a09f4ad1ab33607d2671c5de72ed528b58a76a8" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-16384/natural-quality/metrics-after.txt", "bytes": 2987, "sha256": "eef380c0956b30c89fa1dbc4bbbd7c10718183a946c016145178fd37620588e4" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-16384/natural-quality/metrics-before.txt", "bytes": 2975, "sha256": "d7fe101fce952994e7103dafdeabab8688317789652e3164e1a119448d169e4a" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-16384/natural-quality/output.txt", "bytes": 721, "sha256": "c2145336b5e23b0b9565bfd91e5af45987549f9bb68063eafe50cf778a9acf88" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-16384/natural-quality/receipt.json", "bytes": 8580, "sha256": "30b9f101d5a895e9c1b61787b2501bc8806edeae3cb7bc2ba21b3668d2437659" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-28672/natural-quality/metrics-after.txt", "bytes": 2991, "sha256": "72ec1a1e46190b345f23b50c3062b25dc246fd3a232655bd724eb7850272206b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-28672/natural-quality/metrics-before.txt", "bytes": 2975, "sha256": "ec3f79ec9f0bef85afd840c2c92e90e668a11957f404210e1b11a3f5cc2fda31" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-28672/natural-quality/output.txt", "bytes": 583, "sha256": "ef987712107c63784581a3c3f943a2f879144f17b4d09105aa4233a9c46f486f" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-28672/natural-quality/receipt.json", "bytes": 10520, "sha256": "6cb070fb985c92b6589f8d298f2cf77ec2e4b0010f0e22814bfc926fc3c170db" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-8192/natural-quality/metrics-after.txt", "bytes": 2986, "sha256": "3d154b5ff3abc2e42e062e39018addb90a6a4e7a852f2d1b18f00f606d1f15e7" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-8192/natural-quality/metrics-before.txt", "bytes": 2977, "sha256": "014046b46b4c8ddc5273f5253476ca0698b7022d3baeb975628329cc519130d8" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-8192/natural-quality/output.txt", "bytes": 984, "sha256": "34d57c2c67f31445813be740d59b95d020676e0ef8f1dd78e7fbf6c3ea1d2efa" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/quality-8192/natural-quality/receipt.json", "bytes": 7421, "sha256": "30e94899fee2ac33f63c656c8db51d76cf116d8c2f3ad75aca9350f9b58b175b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2977, "sha256": "36b95112fb87877b4fec284670755cc0c65b502aa366835689312133df2bc317" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/fresh-prompt/output.txt", "bytes": 1452, "sha256": "7920c1c706022e42bc6a4717db769db784fa7e90e4fe2509570164195ea6dd6e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/fresh-prompt/receipt.json", "bytes": 8713, "sha256": "a97286ef07b2a4e22b6f10c8d7cc739770d236bce5f5bb84a01b24f843cbe353" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2983, "sha256": "f4e1cf1bf7cc229355fc00850a4aed45977c23b348992ef3970c5346729686ed" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2965, "sha256": "cc10f33ea335298b02812889211aa700d9d8701f32f8f1789a55229c78f709ed" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/reuse-requested/output.txt", "bytes": 1452, "sha256": "7920c1c706022e42bc6a4717db769db784fa7e90e4fe2509570164195ea6dd6e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-0/reuse-requested/receipt.json", "bytes": 7168, "sha256": "ccad1591daef9215a3541c4b2639972a9ccdfeb7c7ee0e234a21e31246e74a82" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2985, "sha256": "a32f07e2ff296910a29904c59aa918abf06d81a262a73145605490abf877a548" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2972, "sha256": "147df2b55af7808976b27bccfb1df98c73a65c972b82122aab343b7bc5513e2a" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/fresh-prompt/output.txt", "bytes": 1452, "sha256": "7920c1c706022e42bc6a4717db769db784fa7e90e4fe2509570164195ea6dd6e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/fresh-prompt/receipt.json", "bytes": 8710, "sha256": "ef46f1fcb26019f90b558794eaee601d2c877b1cdd1666840c245c3517e71293" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2987, "sha256": "8cc7c46935df643a5e27729b583386645ac4a14974d6de872cbf4a4b971be502" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2973, "sha256": "bf233f5bd0a1dc3b3af7aa0ee3d4499817ebba2da994a896b2b8c2b15aaa035e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/reuse-requested/output.txt", "bytes": 1452, "sha256": "7920c1c706022e42bc6a4717db769db784fa7e90e4fe2509570164195ea6dd6e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-1/reuse-requested/receipt.json", "bytes": 7170, "sha256": "38f8e12ea17f59119fd7df4cdd3aadee20718c91cf43bcde64280edde260df46" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2987, "sha256": "d65f53f85f7f9fa8a5572afee6bab79f0f52e903a27cc38dc3807b05dc8a98e6" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2975, "sha256": "7eb3000dc504e6a18d65e04a36f3eff4f53b783eb36eb24c02a626954b7b99cc" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/fresh-prompt/output.txt", "bytes": 1452, "sha256": "7920c1c706022e42bc6a4717db769db784fa7e90e4fe2509570164195ea6dd6e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/fresh-prompt/receipt.json", "bytes": 8719, "sha256": "4841ffbb058d9ec6ab874f1c067f12f00c4959dcceead216cba051da9b89cb1d" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2989, "sha256": "5142e0818fbdf528d0394115827c856824bfe6896d23b8fc2b6059bcf9ca9f90" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2975, "sha256": "0ef2942ad91b6ee062128a605d3a0716fdc7caaf6fa4ec9939f3570b0c8b7feb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/reuse-requested/output.txt", "bytes": 1452, "sha256": "7920c1c706022e42bc6a4717db769db784fa7e90e4fe2509570164195ea6dd6e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/repeat-2/reuse-requested/receipt.json", "bytes": 7169, "sha256": "c64daa23f02bbc086577c17a9d63aac186f9373472748d661c71341896be99e5" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/server.log", "bytes": 76641, "sha256": "74d57fe640c437e4c0c4876bfc4207b075555ea733798ed50073a5e30dd925db" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-f16/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/cache.json", "bytes": 755, "sha256": "de8b9224cbaa7673c6f1844d094e965755d178b4b45c590477f914732885575c" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/configuration.json", "bytes": 2173, "sha256": "be9f2454261d69f8c89d06cf18b9dd83fa40920bffedc1a3e52e5546c8ab2bb9" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-16384/natural-quality/metrics-after.txt", "bytes": 2990, "sha256": "f199daf0315479a19b7270a25c91db43c0e7a7c1ceb7a857294d59b2995eee21" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-16384/natural-quality/metrics-before.txt", "bytes": 2976, "sha256": "6b5d3cbabc10c51d11e676a06dc7910790b6252c2e06c46121fd8ce995df1c44" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-16384/natural-quality/output.txt", "bytes": 1249, "sha256": "574fe6683a8a47dcef6b2bb575bd86400bc9450a4ab6d1f050ea6831da59248a" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-16384/natural-quality/receipt.json", "bytes": 9359, "sha256": "02b0a3818d52596412e8daf4cc212c0ce1cf0d032a9dfefb63a5a035849cab07" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-28672/natural-quality/metrics-after.txt", "bytes": 2990, "sha256": "b736bfb7db16ab036c0a905ecf4086ff7411b5697313375469e2465cf3b3de35" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-28672/natural-quality/metrics-before.txt", "bytes": 2978, "sha256": "32bbdcf64d9bae5728831bc3b3bb1f6e8475da78f375119f5b43a5ac2e6ec4cc" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-28672/natural-quality/output.txt", "bytes": 728, "sha256": "76186bd4f0156303aed09deb56b7dc776ae6b3bf2ab1b72725e828a6b0f7d924" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-28672/natural-quality/receipt.json", "bytes": 10902, "sha256": "443e757345d7367a88d651f2d2fd520aa98521be6bbedb73bf579071788ad2cc" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-8192/natural-quality/metrics-after.txt", "bytes": 2988, "sha256": "20e69c5185e5f392ef29664f2d403dd3147ef8013a6ae0cac33315d840dc3239" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-8192/natural-quality/metrics-before.txt", "bytes": 2977, "sha256": "5b6db008fad906cab96730984af1c2b3552f4c0a4304e85bf6886f89ea722267" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-8192/natural-quality/output.txt", "bytes": 526, "sha256": "a50d2923e159c2520f0bab50d564a2299f48db305de3a595df7ef435601c8e1f" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/quality-8192/natural-quality/receipt.json", "bytes": 6649, "sha256": "46c45eff5110b3e788719feb371030ebe7183beb70c63716fdbdd7ceac071052" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2978, "sha256": "67e9a8c548b70b59a2cc2676919ce16111c4f3a6afea6ad638b289f2bb0892a1" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/fresh-prompt/output.txt", "bytes": 1580, "sha256": "5b95c87ba7c84b98ed0931455bf6d8ac85f867e49d4e7bbf3ef23139770c749b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/fresh-prompt/receipt.json", "bytes": 9487, "sha256": "21b3cac68fcc21a00c968137185d2141bd99e5b1c7585cf834976fd4d8dcf875" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2984, "sha256": "cadd1f62f160f60a6ca00cd7401f0f9a69698b5b4fe6167505ee8e95ed1fbabd" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2966, "sha256": "4db77f8d9e19553de36fa30d6c59d8ac69b798801f8dd9c289144d596acea6d0" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/reuse-requested/output.txt", "bytes": 1580, "sha256": "5b95c87ba7c84b98ed0931455bf6d8ac85f867e49d4e7bbf3ef23139770c749b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-0/reuse-requested/receipt.json", "bytes": 7549, "sha256": "373904ff57d0fd60b397a60e63473986ba2ef0b6f05b915901b4aca432192a9f" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2985, "sha256": "bf288c877437fb4aa42f766d4823899e8ddfba8ab18255d3e33b28380f3300aa" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2972, "sha256": "c28be6f1ac665d8ca0555677f359b77dbcdc60ca1b7b94dfac3cf4a3313d5618" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/fresh-prompt/output.txt", "bytes": 1580, "sha256": "5b95c87ba7c84b98ed0931455bf6d8ac85f867e49d4e7bbf3ef23139770c749b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/fresh-prompt/receipt.json", "bytes": 8715, "sha256": "40340487cbbf578396aebfdfd153f3599a2b34b35d6b19b5c8547dbdde71c6de" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2987, "sha256": "96d528658cf6fd0665d24701260cb0c9fddfad1c200c761e3ce9581868da9b66" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2973, "sha256": "ac828d57c95423f4addf060397e9fe41a3e39d9b51c13f376e7fa156db4c4ff1" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/reuse-requested/output.txt", "bytes": 1580, "sha256": "5b95c87ba7c84b98ed0931455bf6d8ac85f867e49d4e7bbf3ef23139770c749b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-1/reuse-requested/receipt.json", "bytes": 7554, "sha256": "727301fd79ca2fb5f20889041cb3107915898730729a7f125b4eaa0f2fac6aea" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2988, "sha256": "16d2134afa9460c81d140eaae1d525dadcb8e10fefb770ee74c8e5733aa1e220" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2975, "sha256": "7f2e0dd49a50b1211f7fd3b84bea9ecc886679ca3cf3f23d6db698c35ee1f35b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/fresh-prompt/output.txt", "bytes": 1580, "sha256": "5b95c87ba7c84b98ed0931455bf6d8ac85f867e49d4e7bbf3ef23139770c749b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/fresh-prompt/receipt.json", "bytes": 8713, "sha256": "2aaeced0cef8b3f99f4e5c2a3af1da9cac356c45c1d81adea455a617ae68af09" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2989, "sha256": "55c2d68ed6c3bcc78e158672938cd5f35011462dade06e76ac07d88bd7be6792" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2976, "sha256": "75f962d7c820d786e94d94b732233582662d51508dbbbfd23502b4db099ec698" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/reuse-requested/output.txt", "bytes": 1580, "sha256": "5b95c87ba7c84b98ed0931455bf6d8ac85f867e49d4e7bbf3ef23139770c749b" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/repeat-2/reuse-requested/receipt.json", "bytes": 7553, "sha256": "ea781ddbe2bec014ca3e929e1a48a5973c4b82551304be01b55824372fc3599e" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/server.log", "bytes": 76769, "sha256": "1b046bcf89231ac6fd3b1f0e716aa4c97415a21e01b29c4fac0b4f1e51c3de40" }, { "path": "prefill-kv175-20261005/pinned-hybrid43-mtp3-q8/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-kv175-20261005/requests.csv", "bytes": 2743, "sha256": "0c21761f4bf7a7829114cba8990a3e8438e54c2c3afce830d777ed507f9e0896" }, { "path": "prefill-kv175-20261005/requests.json", "bytes": 158784, "sha256": "1de4cc7c63ae9f30047fad5221a4d9c5df7872c48d36ebc166fde79fa31bb8b3" }, { "path": "prefill-kv175-20261005/resources.jsonl", "bytes": 695925, "sha256": "2632119452bcc7f56dd03aa38714c1b4c4e6d2a94a38944a8d4c1f0d40da8a55" }, { "path": "prefill-kv175-20261005/summary.json", "bytes": 11420, "sha256": "38c17fbfad6e801ca62438be32f0631da68d601dfc23cea5cb5b6ccddbf705d8" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/cache.json", "bytes": 755, "sha256": "a186e69080f1be08507b4b630aa240e9d0a8b47661d3da839f1fcb8d9c94820a" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/configuration.json", "bytes": 2171, "sha256": "6c6690ef429d3cfea4c6be1478e6cd4d845910b0011fa79555987b5cae772fd8" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-16384/input.json", "bytes": 109746, "sha256": "c0eb158a41dfd6832b74ebd791c77b1f96250e57280e9ffee05bb28edabd1214" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-16384/natural-quality/metrics-after.txt", "bytes": 2845, "sha256": "afc380a24f1a026e5aa733f6c02a766c13025116bf98023dd9e9850bb5265737" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-16384/natural-quality/metrics-before.txt", "bytes": 2830, "sha256": "128f49e2add23d50d14fa325cf1cddf173c6864152b29cff20421f510e79c22e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-16384/natural-quality/output.txt", "bytes": 671, "sha256": "142fdcf22c91ed8211a558ade1aa31458832159111cd42ace4938346a0f70db6" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-16384/natural-quality/payload.json", "bytes": 142745, "sha256": "7f7b8f0552565dfe3712f2871a8ff29ce01323b4cdfa639ecd0f3e07addc9625" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-16384/natural-quality/receipt.json", "bytes": 8237, "sha256": "a4159672b25d0d56f76774f85487fd2b33bf95fac6b38660d45837e9d118f5a9" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-28672/input.json", "bytes": 191950, "sha256": "b4a636c058dd0de4ca0407a6af420c0e54c31db8266a4711a8f7a06c1cf86c3e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-28672/natural-quality/metrics-after.txt", "bytes": 2845, "sha256": "c668fd481d617fda2b98c48c602afeeb1683e4e588e973761cac245a3c764d2d" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-28672/natural-quality/metrics-before.txt", "bytes": 2833, "sha256": "f76d390d698777a2a2ba662912e955a85a12677de20b457336e222d38835905f" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-28672/natural-quality/output.txt", "bytes": 704, "sha256": "ed32a88509176be80ac90164c5dfbb1164cc6893e4afcbb74b6a9636e41b4214" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-28672/natural-quality/payload.json", "bytes": 249525, "sha256": "562caadd4c67004f0cdcae4bb6d8d251109dfd1faf13cc41c71f11fc07587c18" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-28672/natural-quality/receipt.json", "bytes": 10562, "sha256": "1a49f3430ea0aac3c6e131b33ef2b9c414a1500a6c6f59c8a40eeaea1af7ea08" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-8192/natural-quality/metrics-after.txt", "bytes": 2842, "sha256": "5bc650b5d16d6678c12da6eca1cbc0e08d9b4331a99f0a3054eb0a3cd11df7cd" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-8192/natural-quality/metrics-before.txt", "bytes": 2832, "sha256": "17f8ecd860b2bd7771b11fd58a3a226851e7e3990301fa54a90e0d3ed5867276" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-8192/natural-quality/output.txt", "bytes": 577, "sha256": "53888df56e8c0f284fea3cf54709788954441b98ca4e0ec6c42c7ebcef7ec7c2" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/quality-8192/natural-quality/receipt.json", "bytes": 6677, "sha256": "bef1a869622d41be01db7b2bea3c6d5490279083bd927fc039a3417ee44afb7c" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2832, "sha256": "bf3275e182fb09d37f52ac5863a4ba5da598bfa80e0223818a19c0fb83662afc" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/fresh-prompt/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/fresh-prompt/receipt.json", "bytes": 8370, "sha256": "cf949735c240452d952ed1d02a435c9f88e52a7f27a6bcf1c5c9e6b8825f982b" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2837, "sha256": "78270047add4a021f3f835aee9d3f647a15811dfc816d52e7ae7d4904fc53ce1" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2821, "sha256": "ab0602bbe74bdf59cee08de1cd51793e8b14da645acd6221b5a8332cfda43d9e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/reuse-requested/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-0/reuse-requested/receipt.json", "bytes": 6818, "sha256": "dc990cd8ee26af00069392806fe77a8b402b8ceeec0388b53b2c291084eb7f7b" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2838, "sha256": "e14050a892592d63bd5d081ebcfb91cfacad39666a86fee93cad7ec36f330af0" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2825, "sha256": "c12eb07bdfab874066f9a7b9f605d8c2869e895db9943e927670b52d25e80b1b" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/fresh-prompt/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/fresh-prompt/receipt.json", "bytes": 7980, "sha256": "f13df9d0157fe899b9f9923233ef4a64ddb6371b7b3014c715f1f011f0c9dae6" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2842, "sha256": "43510bade7d5667ed48930a60b718f96b0274b7d0ecbdd0c72499a4c6913fb24" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2826, "sha256": "843ccfee136858bba8a48869eeaed1e290fe24f1aef7191ad5d2ea7e109f3068" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/reuse-requested/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-1/reuse-requested/receipt.json", "bytes": 6818, "sha256": "e544c524fc6c53d68201eb1afed02f60c8f6c99c52a9600748a58929c2e07b8c" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2843, "sha256": "e7c7bd1636df5d17c6b8b558748c9b8222611ab7af147563351d91e9dbf6f7d6" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2830, "sha256": "4c73c2d7fdf658e3794c66004e6d9e8a668d39b5a3a8c7ffaa0ddf699f81b529" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/fresh-prompt/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/fresh-prompt/receipt.json", "bytes": 7978, "sha256": "71a2c5722441aa145d75a0e4506c1dc613f8b3b0030c65c78b0c254bf5a993c9" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2844, "sha256": "560c0bb5a1d00ca09cdd5af940b6641a320bba74829102d7b52897d8f05a72c6" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2832, "sha256": "22e4c99e3ee7c32d8d9af8763a48f9979bee4ce2cda8e0e7409ad68ac307006b" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/reuse-requested/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-2/reuse-requested/receipt.json", "bytes": 6817, "sha256": "dc07638e95478aed8b36fed26d669c080f9869efdcc31550fcdca3195509ff6e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/fresh-prompt/metrics-after.txt", "bytes": 2843, "sha256": "1893c3630e5165a3c80d290ddd8c8e71a906311ac852ea4af0253b9afca27808" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/fresh-prompt/metrics-before.txt", "bytes": 2832, "sha256": "bf0bc5a735e37e696bb3cf2f14b2a1be3f34502631a9dc0702cf5fad92d80ef9" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/fresh-prompt/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/fresh-prompt/receipt.json", "bytes": 7979, "sha256": "0a1944119672a4127168d3e20a827b7dee955d65a597213f5bfac00a604a5380" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/reuse-requested/metrics-after.txt", "bytes": 2844, "sha256": "1d6707062658884ec71915d4655369f2d57ff0461e6c8fe87b2c0ac4b6c6172e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/reuse-requested/metrics-before.txt", "bytes": 2831, "sha256": "51a39df9d01506eee60bcea04794792db07e2dc2221017b88fe4351018206302" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/reuse-requested/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-3/reuse-requested/receipt.json", "bytes": 6816, "sha256": "554a3e0c92cbd98cd7c148d2b3d1d402ff3c69a7916bb38486b524a6243160c3" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/fresh-prompt/metrics-after.txt", "bytes": 2843, "sha256": "2dba752920c9452f60ca477fc44278304fec5c18d9e69d320f6b1ac397f32fa7" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/fresh-prompt/metrics-before.txt", "bytes": 2832, "sha256": "69e88a81809450b2c71beeca91dbe2ab8cbda354d87627d464a833ecba724ca7" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/fresh-prompt/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/fresh-prompt/receipt.json", "bytes": 7985, "sha256": "867f2d4973d52eaae33570d4cc1b8e7079c6282e00246bbd915896772cce95bb" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/reuse-requested/metrics-after.txt", "bytes": 2844, "sha256": "e04b6d04ab6ceada17bb0582007fbf894ef08347c1acd29dbfffa04b8798c417" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/reuse-requested/metrics-before.txt", "bytes": 2831, "sha256": "5e257bee8e704a6dd576629aab83ed9417a2f50ba8caf9c3d4b378a409a738ab" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/reuse-requested/output.txt", "bytes": 1333, "sha256": "9199fcea18356033cdd3196da8b5ed6b753744b054d684355eb1f5525ef2238e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/repeat-4/reuse-requested/receipt.json", "bytes": 6816, "sha256": "cdb8331eb12a52ad01935a2fba034474e8ecbe420b20955c7ce900970875a48e" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/server.log", "bytes": 91251, "sha256": "d5ae57eaece03ecb62d70b2cc0d9390ac9d211678f15ef53600b982a69fb99df" }, { "path": "prefill-mtp1-finalist175-20261005/pinned-hybrid43-mtp1-f16/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "prefill-mtp1-finalist175-20261005/requests.csv", "bytes": 1996, "sha256": "1bbe81f1caad953eb0fada99962ba7d37d34d401a42b298ad4fd301a1513653e" }, { "path": "prefill-mtp1-finalist175-20261005/requests.json", "bytes": 105071, "sha256": "ae2328e2071fbaade656c1b62a06637baf89b1dcf7039645a8c253eb0b6844e1" }, { "path": "prefill-mtp1-finalist175-20261005/resources.jsonl", "bytes": 466896, "sha256": "0f9008dc93198682096fe0f935406997c2d46fb13e06de6d4d6488a56f195b74" }, { "path": "prefill-mtp1-finalist175-20261005/summary.json", "bytes": 8570, "sha256": "60c449a49076ec96ec75a1567092629cce343d61e0f421f4d616aa3d9c920863" }, { "path": "prefill-observe175-20261005/benchmark/context-32768.csv", "bytes": 737, "sha256": "99ed71cad2e7fa3325f465f06b7f09c6b17d14100426f2f24106ac772e8839d7" }, { "path": "prefill-observe175-20261005/benchmark/context-32768.jsonl", "bytes": 12459, "sha256": "2ba68e02684518163fcf67ff0d66d96a5d620a8a4440facf5bf32a0703ac3e17" }, { "path": "prefill-path175-20261005/requests.csv", "bytes": 920, "sha256": "39c42b0ef0ec0e819870ce2178ebf083cfaff57e0c383a84ca7415cf3f73b22a" }, { "path": "prefill-path175-20261005/requests.json", "bytes": 54985, "sha256": "383644571564556eb11f9042a5b3488716d0e4516bcb2df5f3e67ac487a25795" }, { "path": "prefill-path175-20261005/resources.jsonl", "bytes": 287299, "sha256": "72f3170cafb62e08432bf154c0fa5ef055122325428e3c4e901d436ff3ea5dcc" }, { "path": "prefill-path175-20261005/summary.json", "bytes": 7772, "sha256": "90e3225f94522e357d6f5d294b06ba284f2dc3b5688076ff24be36c6fd5fa7c6" }, { "path": "prefill-path175-20261005/uncached-cpu99/cache.json", "bytes": 837, "sha256": "8ebdbe417981cd3ab6fa1281a126b031ff7f466495b9e815e6adde068a2331ca" }, { "path": "prefill-path175-20261005/uncached-cpu99/configuration.json", "bytes": 1649, "sha256": "1f8498aaa5c872a416120bd1dbd5ba24f7b701126233813fa09b7deb5642f35c" }, { "path": "prefill-path175-20261005/uncached-cpu99/quality-8192/input.json", "bytes": 54925, "sha256": "e704da992f2b8b1b7ca30a59475bcce83c99bb71b19cd847ecb0fe3073c99f20" }, { "path": "prefill-path175-20261005/uncached-cpu99/quality-8192/natural-quality/metrics-after.txt", "bytes": 2594, "sha256": "b2b32323898978c6cb1195cf414d0aa808e71fe56e93fb04859f7241b883c939" }, { "path": "prefill-path175-20261005/uncached-cpu99/quality-8192/natural-quality/metrics-before.txt", "bytes": 2582, "sha256": "8623e4f40afa48b68164aa5cf2fae41fbec724f9355d84c0565cede55146049c" }, { "path": "prefill-path175-20261005/uncached-cpu99/quality-8192/natural-quality/output.txt", "bytes": 669, "sha256": "090493a8d135613f1b291e106b9c981360e4f35bfd65545a2f88763b4843ad02" }, { "path": "prefill-path175-20261005/uncached-cpu99/quality-8192/natural-quality/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "prefill-path175-20261005/uncached-cpu99/quality-8192/natural-quality/receipt.json", "bytes": 6530, "sha256": "c8f479c51a0cfe19aea77af4baf8afde76cc5eb6f99dde0702f617e5385168be" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/fresh-prompt/metrics-after.txt", "bytes": 2587, "sha256": "9389eed121ee106f5972f6675932d628c87f75483d1d91d6f410340e5420bfc8" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/fresh-prompt/metrics-before.txt", "bytes": 2553, "sha256": "033056262d2dfed7ee1a6103733a5e445b13fb53daaa6de97304ae73a95d0f36" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/fresh-prompt/output.txt", "bytes": 1290, "sha256": "d58331d941137951144386d4c8872cc22ddc67643d83a45f21243ba3c8f25120" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/fresh-prompt/receipt.json", "bytes": 8587, "sha256": "ce5d6b2f667593fcbcdfa786a6dc0c15316741d4943883dd4d775c4c5a626bf0" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/reuse-requested/metrics-after.txt", "bytes": 2592, "sha256": "948044798ba5cfe32f1a540f26ac611ada382091e86ac7a29e05d3a014e75731" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/reuse-requested/metrics-before.txt", "bytes": 2575, "sha256": "f7f94605ac91620134857d0ab6caa79c4d212d5d12bae7921c1f04d5353645ea" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/reuse-requested/output.txt", "bytes": 1290, "sha256": "d58331d941137951144386d4c8872cc22ddc67643d83a45f21243ba3c8f25120" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-0/reuse-requested/receipt.json", "bytes": 6656, "sha256": "44b76bb286ccc85a77cd9505520889c31c132c25d502fb0410e08a6309911afe" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/fresh-prompt/metrics-after.txt", "bytes": 2593, "sha256": "27f2391ad75b0b2c94bd281cd843d5b065a161f4b2da0e13e9282bef2b62ccfe" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/fresh-prompt/metrics-before.txt", "bytes": 2580, "sha256": "e934e6d81801ba5c39d83c6172ef337c4a29cdda2a1d0186afc5a67b001bcab1" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/fresh-prompt/output.txt", "bytes": 1290, "sha256": "d58331d941137951144386d4c8872cc22ddc67643d83a45f21243ba3c8f25120" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/fresh-prompt/receipt.json", "bytes": 8204, "sha256": "66618c727692496ff332661f559de5beb0f750e0bcab112aed31ac5f6904aa18" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "f7fb3317fd28d835c71b43212aa8e72ac660b4357c2fd5d46ef6f32ef77d15a7" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/reuse-requested/metrics-before.txt", "bytes": 2581, "sha256": "7d95fe5dc51114fe545a186418082f16f448e4fbf7eae1690b0630f64dbfbd04" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/reuse-requested/output.txt", "bytes": 1290, "sha256": "d58331d941137951144386d4c8872cc22ddc67643d83a45f21243ba3c8f25120" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-1/reuse-requested/receipt.json", "bytes": 7043, "sha256": "71bc521864165bfc2c53d40cc76fcb4d1c7d15c05162ad44e537baed82db5364" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/fresh-prompt/metrics-after.txt", "bytes": 2594, "sha256": "09a56c036b55466547955f583ff90a5f41988dfacae1fa89b02b9897030da5b0" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/fresh-prompt/metrics-before.txt", "bytes": 2582, "sha256": "68ec5f03f30d1bcf70f10d40b8920c9efd69700f36a9c955ce4c9d0deb145d51" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/fresh-prompt/output.txt", "bytes": 1290, "sha256": "d58331d941137951144386d4c8872cc22ddc67643d83a45f21243ba3c8f25120" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/fresh-prompt/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/fresh-prompt/receipt.json", "bytes": 8207, "sha256": "c7009e55e337c2d8fbc9cd393c5ecf12023fb0b43458769886c757e199e81e75" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/reuse-requested/metrics-after.txt", "bytes": 2594, "sha256": "5ab3bbddc34fdc33337fd67586251728a4a11fcae559a35a96f6b35d3a3d8fcf" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/reuse-requested/metrics-before.txt", "bytes": 2582, "sha256": "410fdc4932bc195d52313556f61ef1c4b01d0ae5cbce4ef14adb15bcaa7e4ee0" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/reuse-requested/output.txt", "bytes": 1290, "sha256": "d58331d941137951144386d4c8872cc22ddc67643d83a45f21243ba3c8f25120" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/reuse-requested/payload.json", "bytes": 71570, "sha256": "061bb1ebb55029b3c9c4540e0dbc8ad79605a639ff653c4f9affb53328a87ecb" }, { "path": "prefill-path175-20261005/uncached-cpu99/repeat-2/reuse-requested/receipt.json", "bytes": 7044, "sha256": "d06681a0d4bf5021fb342400e7f019d49e17deea2a0b50e2d8ada1db54c216ba" }, { "path": "prefill-path175-20261005/uncached-cpu99/server.json", "bytes": 355, "sha256": "b7e76fb359939024a9ad0c96979cb055d974e6073c4b6ef1b295215972fc1b49" }, { "path": "prefill-path175-20261005/uncached-cpu99/server.log", "bytes": 48230, "sha256": "bf43f53edbe97489799139324b441da0f4c65717de62f5fac89d055019b404d9" }, { "path": "prefill-path175-20261005/uncached-cpu99/speed-input.json", "bytes": 54935, "sha256": "be6a692b1f38420bcb79688a5633a2ed2d6844257b5c2eb0e1918c57a73c05a3" }, { "path": "upstream178/results/control32-noMTP/quality-28672/output.txt", "bytes": 690, "sha256": "f97c8543b0bde398dcfdbede0453b617a4824f9c9e302be77eb48e010515af49" }, { "path": "upstream178/results/control32-noMTP/quality-28672/payload.json", "bytes": 249525, "sha256": "3cba61a6792cf8982286162eb252d55c2bb51c905e30dcbfa7622c3169c3c917" }, { "path": "upstream178/results/control32-noMTP/quality-28672/receipt.json", "bytes": 9178, "sha256": "003d0ce32cf4c9d1770b7734525a2b454b1bb0436e1c6937653df3cacdf59af2" }, { "path": "upstream178/results/control32-noMTP/quality-8192/output.txt", "bytes": 559, "sha256": "dc5e291c758de85439182809eadc9670b8f96df9fa5c67085b32bff4711ce1d4" }, { "path": "upstream178/results/control32-noMTP/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "upstream178/results/control32-noMTP/quality-8192/receipt.json", "bytes": 5299, "sha256": "f5418042a6b550efbfde3d404f02af658f6b1a6fc197885ac8074b54837ca0a0" }, { "path": "upstream178/results/control32-noMTP/requests.csv", "bytes": 766, "sha256": "fe3dfe59e5f2707aa0ab0453b05ad0efa96574410831b8d410b34eefd10283df" }, { "path": "upstream178/results/control32-noMTP/requests.json", "bytes": 34778, "sha256": "41ee43a7a40ec41fd35382ef0280d587483484892350d003f9d4072c74cd3db0" }, { "path": "upstream178/results/control32-noMTP/resources.jsonl", "bytes": 262079, "sha256": "24ef95e789e35d0bb6f78c7023d25630fbde779fd07b0b390b84ad76fa13e43e" }, { "path": "upstream178/results/control32-noMTP/server.log", "bytes": 38293, "sha256": "e83ea820a22b9f0d74baa259fb886e0061715140cdba1db929d2822b7fa4a627" }, { "path": "upstream178/results/control32-noMTP/speed-28672-0/output.txt", "bytes": 1232, "sha256": "64197c3bc8f8b653988f77da1ade71dc5fe1398682112ab9077786812478bb12" }, { "path": "upstream178/results/control32-noMTP/speed-28672-0/payload.json", "bytes": 249557, "sha256": "19766cf4c14b17729e8087d3260c4a66034d819410f62423c9f721a6af965b3b" }, { "path": "upstream178/results/control32-noMTP/speed-28672-0/receipt.json", "bytes": 10876, "sha256": "07916571f0ec9f2befdcf6179455af0f2f9e41fc2fdcb57e6484d085cc0d64b8" }, { "path": "upstream178/results/control32-noMTP/speed-8192-0/output.txt", "bytes": 1210, "sha256": "60f6b0aca7690bf7c0d54a8f648560734aa6d9e08bd3a5fc637dffec771d7628" }, { "path": "upstream178/results/control32-noMTP/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "upstream178/results/control32-noMTP/speed-8192-0/receipt.json", "bytes": 6994, "sha256": "c4db5a3bd97e9a7f509c693b7a4faabf3a6b0c54de948128b59008fcd94889ab" }, { "path": "upstream178/results/control32-noMTP/summary.json", "bytes": 6496, "sha256": "5c0debeab732d308cd55543ee0d9dd135d76dde4915bbcd1eedf0b36f47cb23f" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-131072/output.txt", "bytes": 818, "sha256": "1e35e07c753f8d600b1eea139a37c4e9c54c30c5fea20bb6eae7167e08f76c0a" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-131072/payload.json", "bytes": 1139515, "sha256": "1496901b3b74ec4ed9169c4dd38dd74cbfc2787ae8236e17b2a7dfca8e878eaa" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-131072/receipt.json", "bytes": 39008, "sha256": "7efc17d81b4c234daf608d3baa23a3df37760ea8cea1c839392d3352b08ec020" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-235929/output.txt", "bytes": 694, "sha256": "96652142b8b10fbffcd2d878213abecf1dc5c114215836f530297b62669f52bd" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-235929/payload.json", "bytes": 2050818, "sha256": "8a53e56f2ed23b00dfb1e6ac4046f9e652c55181d5d2a0156684a124eab8a32b" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-235929/receipt.json", "bytes": 70091, "sha256": "9c62940474a593935f8c4711b316dd2a7adbc0c100a42357e00549a176e352f9" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-8192/output.txt", "bytes": 437, "sha256": "dd23c6d31a5440b5240893dd092b06369646f153ae412d87d3e1cbfecd90eaae" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/quality-8192/receipt.json", "bytes": 5703, "sha256": "6ec0f5a0083c95b4f63d36f1651ebfa20804c33e8291d742479378f2ab896b0d" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.csv", "bytes": 1127, "sha256": "19076231d9fd522c41b92dad45c93adb3dd4a87989594c97e4cfac5a1647221e" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/requests.json", "bytes": 251444, "sha256": "e72f832f72636d0b1802d37fba1c410151e8ee39bc506847a3af5205d4b35e4b" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/resources.jsonl", "bytes": 1647545, "sha256": "44f8809932a24b3db22ec85dac75c565dc2b000b58b9c39362036ff0c5cc2350" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/server.log", "bytes": 69755, "sha256": "ea86355e926e5e4c605b689040f1adcc98abae5fa1d6b7cd905557876aefdad6" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-131072-0/output.txt", "bytes": 1272, "sha256": "7a4987e083d17588ae789325f0a02f0faa3bfc1c927c2e1da1c4905042eda493" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-131072-0/payload.json", "bytes": 1139533, "sha256": "2bdfeeb8e20c755b8f214b40a84f3d927ec51ad3fca6b28b25cefd163029ac5e" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-131072-0/receipt.json", "bytes": 39954, "sha256": "eb910604c80b7e56c03a95fc4c6062acd81e004f49324083bcaa84c0b7c6fe2b" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-235929-0/output.txt", "bytes": 1395, "sha256": "8af07cdb3a5bd17d904a87191835c63c25d5e64b9155ac42ec70d67ea8eae242" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-235929-0/payload.json", "bytes": 2050852, "sha256": "e45c84a6cd216fd45768d8b83210d2c270dc3274a79884f13f005a9dfc592fb0" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-235929-0/receipt.json", "bytes": 71341, "sha256": "6379a04b41b7e21f1ba9e2c74b4fe911f4399be9541686dc58df23df70a644bc" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-8192-0/output.txt", "bytes": 1520, "sha256": "49d734ed91128d6174e3cd7b7319355c9039d805ba23094c6cbd87df15911ba5" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/speed-8192-0/receipt.json", "bytes": 7392, "sha256": "d4a935329e9436bb045d324276074cfa386a5cd00cef3bcba9574ebeff75b63e" }, { "path": "upstream178/results/upstream256-cpu43-u2048-noMTP/summary.json", "bytes": 6930, "sha256": "46a8cfacb8e8799cc45e0c8a11a29d64ebef40ef180086b84f77ea781d07df2c" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/quality-235929/output.txt", "bytes": 935, "sha256": "1631893f1cd855731f80cf3280a5db94c3b8aee69126c8a1232cec9fcbcccdb7" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/quality-235929/payload.json", "bytes": 2050818, "sha256": "8a53e56f2ed23b00dfb1e6ac4046f9e652c55181d5d2a0156684a124eab8a32b" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/quality-235929/receipt.json", "bytes": 55321, "sha256": "98bce8edb1de8d5f1814e7a3e2318f59e33eee1777306f1a278ae25c9b17fe3b" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/requests.csv", "bytes": 491, "sha256": "825e07aa8050e09c1f9cef95c09e90a19ec21db48e2f36a0a449a55aad2a0ec8" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/requests.json", "bytes": 120120, "sha256": "3b7049b12e032ae91f9608158dee9267dcb16187833d5bfbb045f6b123421a8b" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/resources.jsonl", "bytes": 853750, "sha256": "9070dd46f40438a572c681de2740bf34b280f7dc32a8182e3c90fea5a3945f22" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/server.log", "bytes": 43453, "sha256": "86167ef804d2010c47540b1e114112d95ef2a1e2d8b0e1d0fe8bc32f7c7fe84d" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/speed-235929-0/output.txt", "bytes": 1559, "sha256": "0118305bf3c7032346981340daa8df50f87f5cd08a500ff6547ac536ae5dccfa" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/speed-235929-0/payload.json", "bytes": 2050852, "sha256": "e45c84a6cd216fd45768d8b83210d2c270dc3274a79884f13f005a9dfc592fb0" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/speed-235929-0/receipt.json", "bytes": 56202, "sha256": "1ca90b6dea9177dbc9b918f8a725d93f440ff8e204ea4dc8048b24abf77b004a" }, { "path": "upstream178/results/upstream256-cpu46-u4096-deep179/summary.json", "bytes": 6373, "sha256": "364bfb1390edb821e2b87e1921af4f340f17d0c7f291d447b793c03437bbdf27" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/quality-8192/output.txt", "bytes": 846, "sha256": "bd19262f04ce9fafd67c2a43d26af4492bcdb980ddda08ef43f405f90be03fd9" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/quality-8192/receipt.json", "bytes": 6081, "sha256": "97dc2f9f8a9275e00fe5d73c7ac38260c107005351a76688e9df92c313c064fe" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/requests.csv", "bytes": 467, "sha256": "2eb5eb211405ace3675f49d3129fd319d983cf0c8ae144bf415325b84f97e1d0" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/requests.json", "bytes": 14476, "sha256": "c10d3cfc6128e73ddea302b574fd15edd45492afe5b368cc83d99703d23ab7e3" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/resources.jsonl", "bytes": 159671, "sha256": "fe246f75ca95ed35c54464ed8421d31baa2c299508a096819b58e0cb02c480bc" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/server.log", "bytes": 29133, "sha256": "41a8d8b57672f980a47df8b1b25e31a8257a74cec20b0768ceff770fcf18abd2" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/speed-8192-0/output.txt", "bytes": 1443, "sha256": "f6dfbb35081d0e1b42b442bd71283963f971e5f8888257f698713cc52a9cada1" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/speed-8192-0/receipt.json", "bytes": 7388, "sha256": "3b206b6730a28f5e4f799de23813a69a011cf0cd4c951d299cce35592915e693" }, { "path": "upstream178/results/upstream256-cpu46-u4096-short179/summary.json", "bytes": 6367, "sha256": "9b0adac2f2dbff4be04fddef2aa90b094caafd2f1c217a6711956c6453044ef0" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-131072/output.txt", "bytes": 1064, "sha256": "4d53ddb11ad914fa72f4902d978111479bbc14baac45e5d47a854a3a97ed3d31" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-131072/payload.json", "bytes": 1139515, "sha256": "1496901b3b74ec4ed9169c4dd38dd74cbfc2787ae8236e17b2a7dfca8e878eaa" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-131072/receipt.json", "bytes": 31673, "sha256": "36bfc9ada0aa36b209817187f267096c0b457fea5e76694c49b848ffee9e195c" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-235929/output.txt", "bytes": 888, "sha256": "33fcc44f5768570238e918b58df15060e2186f44c65d710e6a182efe4cce8cb2" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-235929/payload.json", "bytes": 2050818, "sha256": "8a53e56f2ed23b00dfb1e6ac4046f9e652c55181d5d2a0156684a124eab8a32b" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-235929/receipt.json", "bytes": 55727, "sha256": "733470148686f963c0ab9169068c7af8d7237046e13668c5ec728da19565bff5" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-8192/output.txt", "bytes": 846, "sha256": "bd19262f04ce9fafd67c2a43d26af4492bcdb980ddda08ef43f405f90be03fd9" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "upstream178/results/upstream256-f16-noMTP/quality-8192/receipt.json", "bytes": 6084, "sha256": "19e57bbc8b4f449a08e18307cf490aac9809def9d8a2e5341b77a272e63b9150" }, { "path": "upstream178/results/upstream256-f16-noMTP/requests.csv", "bytes": 1127, "sha256": "e29fedfb7e10ddfce4e096d02704dd7f487ff249843d0ad4363e6656228293e9" }, { "path": "upstream178/results/upstream256-f16-noMTP/requests.json", "bytes": 204710, "sha256": "f31f9bf65582b7cd24d5f2f3d7b367fc9125928859dd468c5fc7c512fd40d3d7" }, { "path": "upstream178/results/upstream256-f16-noMTP/resources.jsonl", "bytes": 1334874, "sha256": "f18b9a79f943eb01c2085c479fe450cedaae0d9a00803e5d91a4666373442ddf" }, { "path": "upstream178/results/upstream256-f16-noMTP/server.log", "bytes": 70199, "sha256": "8d95e05b344ac77e9050348458dcd97adc5be5fd7c377b8e9ee14bf5f300d8f9" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-131072-0/output.txt", "bytes": 1497, "sha256": "d66b11dd1aa74a6932cb23673929e6d31cee33e083ccd5e006453c08b7375d6c" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-131072-0/payload.json", "bytes": 1139533, "sha256": "2bdfeeb8e20c755b8f214b40a84f3d927ec51ad3fca6b28b25cefd163029ac5e" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-131072-0/receipt.json", "bytes": 32217, "sha256": "76a9cc4aa9d578c3cf27a2011f0030bb17a6a26e056f921a73c0755cc616b5eb" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-235929-0/output.txt", "bytes": 1347, "sha256": "43bf3d9c5869e98dae53c6c17b8525b570916210a2582cbb32967122e8809acf" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-235929-0/payload.json", "bytes": 2050852, "sha256": "e45c84a6cd216fd45768d8b83210d2c270dc3274a79884f13f005a9dfc592fb0" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-235929-0/receipt.json", "bytes": 57025, "sha256": "e0c7490eaa78ac5694a5d99a9fda6120a388e95eccde73ac1e8a457759c44789" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-8192-0/output.txt", "bytes": 1344, "sha256": "5debec69ed470bd593ad7929630ce49f34915425e94259f36a22dab8408a79aa" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "upstream178/results/upstream256-f16-noMTP/speed-8192-0/receipt.json", "bytes": 7389, "sha256": "97c05b1e49ae0bb467c052a6d62832c75f8377a370a8b124cbbdbbe1ae0ced13" }, { "path": "upstream178/results/upstream256-f16-noMTP/summary.json", "bytes": 6932, "sha256": "c999cba80d5ade26cfb4b9e93f63295854dd7118644150a8cc7967b10c881c6a" }, { "path": "upstream178/results/upstream256-noMTP/quality-131072/output.txt", "bytes": 1048, "sha256": "e8784a104d02f222f24fb5dcbf14113058015e1c059bdd50743abc7a212de4e3" }, { "path": "upstream178/results/upstream256-noMTP/quality-131072/payload.json", "bytes": 1139515, "sha256": "1496901b3b74ec4ed9169c4dd38dd74cbfc2787ae8236e17b2a7dfca8e878eaa" }, { "path": "upstream178/results/upstream256-noMTP/quality-131072/receipt.json", "bytes": 32411, "sha256": "3d30044f2b7e161cc88e8ffd4816de690050e730fbf0790edeefa318a824a99e" }, { "path": "upstream178/results/upstream256-noMTP/quality-235929/output.txt", "bytes": 692, "sha256": "24082906216043dfa65ab2e3ca118385b64099c4c23faa2f32618c0037c028b7" }, { "path": "upstream178/results/upstream256-noMTP/quality-235929/payload.json", "bytes": 2050818, "sha256": "8a53e56f2ed23b00dfb1e6ac4046f9e652c55181d5d2a0156684a124eab8a32b" }, { "path": "upstream178/results/upstream256-noMTP/quality-235929/receipt.json", "bytes": 56106, "sha256": "2c015ea409f1a3d544707b546f0e51eeec2c8aadd2d7001723bc34c2fd99292a" }, { "path": "upstream178/results/upstream256-noMTP/quality-8192/output.txt", "bytes": 553, "sha256": "fa7937a87eaec3750cd21a2633aabd425e5eb585d8f8368521df80a6ef41e4b7" }, { "path": "upstream178/results/upstream256-noMTP/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "upstream178/results/upstream256-noMTP/quality-8192/receipt.json", "bytes": 5703, "sha256": "fac38a6c651400fc0d55f818b2bbd9f6a57d3c5799ef7885299166ccdb7a0a58" }, { "path": "upstream178/results/upstream256-noMTP/requests.csv", "bytes": 1142, "sha256": "8cbcfd7262458a0c433600effda0b8137502dc772a698c09d6b49f1f50ad1609" }, { "path": "upstream178/results/upstream256-noMTP/requests.json", "bytes": 209174, "sha256": "570040ab12f3644b173b6d1a5f1e215c18e7bc5810af6af75d83b04adf51eb48" }, { "path": "upstream178/results/upstream256-noMTP/resources.jsonl", "bytes": 1363216, "sha256": "87fee35829e6fb0b010f0ab0bd652a745e7a0b44158762c65af9faa55ee5cc0e" }, { "path": "upstream178/results/upstream256-noMTP/server.log", "bytes": 71018, "sha256": "4dfd17a49b829408ed16aa7d97c12533a15cdfce1f7d95556fabbb9cafea4e1a" }, { "path": "upstream178/results/upstream256-noMTP/speed-131072-0/output.txt", "bytes": 1292, "sha256": "de416675d043bb5aa42f13b2da3b869dfac20f93f0ceac37a07048e89004f1a9" }, { "path": "upstream178/results/upstream256-noMTP/speed-131072-0/payload.json", "bytes": 1139533, "sha256": "2bdfeeb8e20c755b8f214b40a84f3d927ec51ad3fca6b28b25cefd163029ac5e" }, { "path": "upstream178/results/upstream256-noMTP/speed-131072-0/receipt.json", "bytes": 33351, "sha256": "f07141eb93bf52568ba32af446ec6e55edb89e21c95c667da5958120abfb0707" }, { "path": "upstream178/results/upstream256-noMTP/speed-235929-0/output.txt", "bytes": 1318, "sha256": "6f4605b6c4fc6b3e2f2a62c169a6543e1cd447fa3359577976563b3f77330da8" }, { "path": "upstream178/results/upstream256-noMTP/speed-235929-0/payload.json", "bytes": 2050852, "sha256": "e45c84a6cd216fd45768d8b83210d2c270dc3274a79884f13f005a9dfc592fb0" }, { "path": "upstream178/results/upstream256-noMTP/speed-235929-0/receipt.json", "bytes": 59286, "sha256": "3698006b6aa91560ef12ba3ee352323b75ba90acb7a9aabdbedd56fd593eac31" }, { "path": "upstream178/results/upstream256-noMTP/speed-8192-0/output.txt", "bytes": 1587, "sha256": "38ecbb13a2bdbcb809f5de11cf6894ef86ffa0b8022eabbaef758ea87e5cd450" }, { "path": "upstream178/results/upstream256-noMTP/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "upstream178/results/upstream256-noMTP/speed-8192-0/receipt.json", "bytes": 7392, "sha256": "13d075161ceee6be3dd337551853641305c8d0863b973ab08ea4925b7f4b708a" }, { "path": "upstream178/results/upstream256-noMTP/summary.json", "bytes": 6933, "sha256": "d6f234b71ffcd674d84abb60a8e9585f00628fe063ee50b2d449d661108e431b" }, { "path": "upstream178/results/upstream32-noMTP/quality-28672/output.txt", "bytes": 600, "sha256": "9ecf826b60fc5c9d3c41281a033d2ee44b5a30b021e4db69a8e673108880636f" }, { "path": "upstream178/results/upstream32-noMTP/quality-28672/payload.json", "bytes": 249525, "sha256": "3cba61a6792cf8982286162eb252d55c2bb51c905e30dcbfa7622c3169c3c917" }, { "path": "upstream178/results/upstream32-noMTP/quality-28672/receipt.json", "bytes": 8409, "sha256": "eb3dcaf20443e43ca1c94f21c333fb6fa04993276ac8a8ed09f36d6ae4d53e52" }, { "path": "upstream178/results/upstream32-noMTP/quality-8192/output.txt", "bytes": 651, "sha256": "b9e217b270401d1da2ba57b70592f3730317d4f34327cdc5e5764eb1a4e33b99" }, { "path": "upstream178/results/upstream32-noMTP/quality-8192/payload.json", "bytes": 71540, "sha256": "16f0fdd9861e038ff47c6a55a814612264fea8b993b6680d625bce6127106dbc" }, { "path": "upstream178/results/upstream32-noMTP/quality-8192/receipt.json", "bytes": 5300, "sha256": "8432830f67f966a456595d818b0ce8be758161f37e6eb2a9ebf888b4578781ad" }, { "path": "upstream178/results/upstream32-noMTP/requests.csv", "bytes": 770, "sha256": "242c560cd1f3dcaf7383a4f387c9353c75b76a39732f3cb7f193709367e98d9f" }, { "path": "upstream178/results/upstream32-noMTP/requests.json", "bytes": 32699, "sha256": "2e478a9a8ef727b7a0d4966bcfde2dd6b9b62df6a4c61de2344ed8ffef55b0f6" }, { "path": "upstream178/results/upstream32-noMTP/resources.jsonl", "bytes": 251328, "sha256": "ae1fd943fcf143fa3739b156ed449bfd5d76e48fdc73a7b3f801bcb26063c0ef" }, { "path": "upstream178/results/upstream32-noMTP/server.log", "bytes": 38223, "sha256": "ca79be249966b951d47e9e5036e05646b53dd8bfbc756832efe7b67ae26e1f1d" }, { "path": "upstream178/results/upstream32-noMTP/speed-28672-0/output.txt", "bytes": 1228, "sha256": "8aa1e5e68d38fa7b2b9a184358d13dd2a4de86b9a98aa3149c1b6c7d67aa9ef1" }, { "path": "upstream178/results/upstream32-noMTP/speed-28672-0/payload.json", "bytes": 249557, "sha256": "19766cf4c14b17729e8087d3260c4a66034d819410f62423c9f721a6af965b3b" }, { "path": "upstream178/results/upstream32-noMTP/speed-28672-0/receipt.json", "bytes": 10102, "sha256": "220d8092d30850024ca0ddd66dcca8405bc6f7cba5e0ce9adfb9d0ae85df2691" }, { "path": "upstream178/results/upstream32-noMTP/speed-8192-0/output.txt", "bytes": 1153, "sha256": "cb84f92c7ae58d7e60eb11be39c9ef55814cfb910cce4680e60086b0f7d50bf5" }, { "path": "upstream178/results/upstream32-noMTP/speed-8192-0/payload.json", "bytes": 71571, "sha256": "a51130c60937874edf1738e8e67d2a5ab6dfc6a709434c21bf3fae7ab0d82f39" }, { "path": "upstream178/results/upstream32-noMTP/speed-8192-0/receipt.json", "bytes": 6607, "sha256": "e6a49f65bba6f20df86fe57383e620b42e00395b994b680e307774094889c328" }, { "path": "upstream178/results/upstream32-noMTP/summary.json", "bytes": 6642, "sha256": "2bdd58a9329b42a8d1800349601c64370732f230d6809f4622dd91d5a33bf7bc" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/bench.stderr.log", "bytes": 181184, "sha256": "38fa3df47dfba689200b919b1ddd4437fab1359a290f68004a23156c2f193bf5" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/provenance.txt", "bytes": 2123, "sha256": "a75b75c63cacb66eb5eaf2fc0b3f1a52323ea4337b17bc0fdd67e264ed453b56" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/receipt.sha256", "bytes": 740, "sha256": "d1227264b40b8037ece4af9c6f135e6371f7ee72dea4de6962d6b05c87c756ff" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/resources.csv", "bytes": 1900, "sha256": "83eea079177539e0f2df9cb8c70d60fdb9327eb6e50c076d7523e521e0216e04" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/results.jsonl", "bytes": 2265, "sha256": "7c7c17e0cd1fa69e131cff29b73a6cb3406e3155080768b3e512fba00b9c67f0" }, { "path": "historical180/benchmarks-20261004/cache-c32-t14-a1-20261004T214318Z-578888/runner-exit.txt", "bytes": 37, "sha256": "64106a8dabc88a2bfaf2e9d8f44ea9b7b8ebf86a8a882b928fbdf032ffdd3658" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/bench.stderr.log", "bytes": 181184, "sha256": "73b9d77403106a8da807cf15dd7fed38481c3329b06f81bcbd1e70ece4c77753" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/provenance.txt", "bytes": 2123, "sha256": "1cc229adab87896c90605bffdbe26a22ad5920446af4cf3ad546adf81eb7b42d" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/receipt.sha256", "bytes": 740, "sha256": "d9048a89ec5873d6d61059f637e5f6aa049503e1143696d321062e0dd3a9348b" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/resources.csv", "bytes": 1835, "sha256": "c2e99e389c24c50d020d2f25183b66f442944e2c6821052e1dfeb6b765d73301" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/results.jsonl", "bytes": 2264, "sha256": "36dd6008ff9c726d1ef1bf218390853667493641c8c0eb89497710de9c556020" }, { "path": "historical180/benchmarks-20261004/cache-c40-t14-a1-20261004T214501Z-581674/runner-exit.txt", "bytes": 37, "sha256": "94ac69b4e6b35b9f4a21dc8bbcd3e8f45d91371bec3a936a89f788712f9f306a" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/bench.stderr.log", "bytes": 181184, "sha256": "cd19752a408ede91cf8cc3b4e80267cf3ba582055a691ded4a46a4e137ab6c8b" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/provenance.txt", "bytes": 2123, "sha256": "bdea10d216c872f6e161212eb146c8e1f87be49194ad90b82403af34971b8e4f" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/receipt.sha256", "bytes": 740, "sha256": "fa8b10254e1c1ce62318b35d0e243313399a7cad122cab2ceb9f6a6100c0a4bb" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/resources.csv", "bytes": 1835, "sha256": "afaa3bd12bd13dee38016701260cff27505dfe93ffc148db78641239606c9687" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/results.jsonl", "bytes": 2265, "sha256": "9a8a4e39214934adff34bc4755b8cdd4ac245f4cd37f3345f795ac4cdc62d2cd" }, { "path": "historical180/benchmarks-20261004/cache-c48-t14-a1-20261004T214653Z-584724/runner-exit.txt", "bytes": 37, "sha256": "45324f64778e9373303d1d5663f4c6f0172e15dfd894afb17ee0d7e6e3967e0b" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/bench.stderr.log", "bytes": 1168820, "sha256": "16236e27ddaeb7204f660bb480b482cf7be51691521f876e9c4c9a31c74070c2" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/provenance.txt", "bytes": 2125, "sha256": "4028072afa973d2cd48dc5db011c0da2873ca854b641766c9e1b076f948d207a" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/receipt.sha256", "bytes": 748, "sha256": "1311b87029dd0efa0c3e86b93e1748ebf3d7350a8b97475b1ed4c8f0b265c63f" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/resources.csv", "bytes": 1699, "sha256": "bddd56771180e934ce0f186af804d1269e64549f7298a9d342f84a45a46e0a0c" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/results.jsonl", "bytes": 2264, "sha256": "51b9db09d6f4c913e4f86a2833a7cddd68cb355e25c9dfd33c8da79ec4faa68f" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a0-20261004T214842Z-587507/runner-exit.txt", "bytes": 37, "sha256": "244d1bec8a25fe7afbc5caf4db4fe5d383e5c1686ea8f19568086305ff91cddf" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/bench.stderr.log", "bytes": 181184, "sha256": "c988525c55d1996ba3ae2b583fa176305881cef4fb9819e64437cd770599ce73" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/provenance.txt", "bytes": 2125, "sha256": "b298e7d289fcf5f9e5f0d43a6ba7565a740f2664e942d0506ad0401057f1178d" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/receipt.sha256", "bytes": 748, "sha256": "bec4a7578733cf395d5a636f3ed970223620c0dd88cfe5d6cc4d4f36fe778617" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/resources.csv", "bytes": 1835, "sha256": "7e50b535a6e3ebd1aef60a9969cfb99384026fdbb02d7b3ca9ee585ff084fa82" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/results.jsonl", "bytes": 2263, "sha256": "dbca6964ce8a5ce3149b84b5c70adb33b614ea59c6e36968f5893a6a02a5c258" }, { "path": "historical180/benchmarks-20261004/overlap-c48-t14-a1-20261004T215031Z-590180/runner-exit.txt", "bytes": 37, "sha256": "c2265275d0bbf5e4b74d7e8e237f63e81a2366405b04be6dc17c82f97be0eece" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/bench.stderr.log", "bytes": 171387, "sha256": "17f2ebaeb694e316caab43434946e0166426dbd85e1acb084d7c55f30909d17f" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/provenance.txt", "bytes": 2560, "sha256": "10eb52c1c889a53ebbf2c780c20a4db623d6ed09709bc78336616aba484d9b28" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/receipt.sha256", "bytes": 812, "sha256": "b5257fa3829b73fa9979fe523dab69e89932c92a134f07589b4a9ef81547f9c6" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/resources.csv", "bytes": 1971, "sha256": "3dc45d9b040344543de2bc55354d6b98362a643fa549179ab5e833f2e5561b07" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/results.jsonl", "bytes": 1137, "sha256": "8889121a50e4975cfc60c415e21f5c8619476775f72a6bf05e21d9af74e724e8" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u1024-pf0-20261004T223734Z-660149/runner-exit.txt", "bytes": 37, "sha256": "c4b832e2c1ccc994fc30fa5933472ad4c8f5f6b93120007ef0a30bf11f6fb2dd" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/bench.stderr.log", "bytes": 171382, "sha256": "15dd7ad658694f1f2a3e38af582710cccc559dd11d8890b5fd21f8fe7326cc1a" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/provenance.txt", "bytes": 2558, "sha256": "60ddd1ff29bbc3f859c6ba9386653f02813d30c39bc8e3285ea7c065504f8b56" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/receipt.sha256", "bytes": 808, "sha256": "53ac722bcca3a8f5be4ead0872f810b4ea833f19af46b0c7281445370b982ee4" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/resources.csv", "bytes": 6593, "sha256": "b0749629eaa1bce49e50f7d83ebb84e849b6e7a00050abef2d4a9f892c57fc64" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/results.jsonl", "bytes": 1136, "sha256": "fa27ba388a0d88ac40739baaea1796a11f01d6083a0b3b9a15973f35f169776d" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u128-pf0-20261004T223048Z-649423/runner-exit.txt", "bytes": 37, "sha256": "dca6f7c6353d7f3631d1084dfab5f5aed5e3c6d089125348f94ce8d61ee31c6f" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/bench.stderr.log", "bytes": 171382, "sha256": "5f1b79d10ddc68e86a6b655407b278b715a92e9d97547035e459bb166d8a383e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/provenance.txt", "bytes": 2558, "sha256": "5e50859a1dda06bbe235fb57b53cb95e66454ea021c70713db9fb1893a31b12c" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/receipt.sha256", "bytes": 808, "sha256": "c72597a3db335fb75722cace7e47bdc18a941ec73fdd50d4624f26229fdfcacc" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/resources.csv", "bytes": 4349, "sha256": "145a4377536d150653a48ab94a14c99ed8bfef28598aba261525c14e6e6774db" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/results.jsonl", "bytes": 1136, "sha256": "5a7e4df43f0b0b56f751acb431f15e6498d5e7eb27a69bd78ba1ae30e0a3d00e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u256-pf0-20261004T223403Z-654270/runner-exit.txt", "bytes": 37, "sha256": "04d8caac40c459ad9900145f2a99ea205bb994d2f55869616231e14f48c9d110" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/bench.stderr.log", "bytes": 171384, "sha256": "e2ff398452ec4678f73fd9fc8d8104ecd149ef3d545b394435fd03e3d5caa0b5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/provenance.txt", "bytes": 2558, "sha256": "83ca1734bde8768fddb14ac16f62346b944830adc10b85ec83f20d664e4d33f8" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/receipt.sha256", "bytes": 808, "sha256": "6675dfc9eaaaf36fbddd9db9b3a8d1055d07a40acdcd79ab07d916882a123b73" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/resources.csv", "bytes": 2855, "sha256": "0dff3ff0063f5e6640200d4134dc7e6314f8757c9d693328253a8bdee4a9805b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/results.jsonl", "bytes": 1137, "sha256": "39888789994ee079760655a6a7003a4e91b57ab865e7bb65b9863bda3b947afc" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b1024-u512-pf0-20261004T223610Z-657788/runner-exit.txt", "bytes": 37, "sha256": "a1b0aa3e3026ee59a3d7ae2c4450d5bf0cf3a74532a5dafea9a5cfcf49b21df2" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/bench.stderr.log", "bytes": 171387, "sha256": "655e4a6537d5e0d939ec06b2d18f90081a522463aee30b79dece4a4570f0fed2" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/provenance.txt", "bytes": 2560, "sha256": "67954303ce91a62534d67d9bb34c85fac95a6d5fd2dc9344f79def7fdebfa1f0" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/receipt.sha256", "bytes": 812, "sha256": "a39eea114c59c9cbef1db7eee902a52421532e8098880e00615448bd70a7323b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/resources.csv", "bytes": 1971, "sha256": "e93b58968fc74f1d1e4af619cfb4eec803147c8ab1d8b2185537f806faec7430" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/results.jsonl", "bytes": 1137, "sha256": "661868b3ce91d08f020e3bae830898d6045fcd588250416e22bbb1ec1d04881c" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf0-20261004T224515Z-669667/runner-exit.txt", "bytes": 37, "sha256": "15dde6cc64dc5b021b3c971ad1b12d73c7a16f3656aa9330a6565044bf9b8ac6" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/bench.stderr.log", "bytes": 171329, "sha256": "2435b228493460710bdb7c128ca546670078bf55daa0a6a4023b0860dabb216f" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/measurement-status.txt", "bytes": 24, "sha256": "2943a6a7c2b1716293e5dc26c5f4fce3334d9b6829d4e08aa506bb067a53e487" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/provenance.txt", "bytes": 2526, "sha256": "ded5755269689995cbfaa4c720f855164ecbb34a966534f372343147d9e0a11c" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/resources.csv", "bytes": 1306, "sha256": "3ace98a28ae5442f94fb00c70aaf5ce8bb73b58aeb4f28b0e3000f5fa2de0980" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/results.jsonl", "bytes": 3221, "sha256": "93736ea0d0dd4828ceb3da78311eac90b6ab9e0bbc581e72ef705729bd0637fc" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u1024-pf2-20261005T060634Z-977668/runner-exit.txt", "bytes": 37, "sha256": "aa6ae3d773a36e818f3bb40e334fb3053ac1e58adc9a23081de38e546f2954aa" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/bench.stderr.log", "bytes": 171382, "sha256": "3de79084672197805f06db123d904c37be2a79d4fa09e9ebbe6e309007a4ffe0" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/provenance.txt", "bytes": 2558, "sha256": "9c5e0797536c61dc31b2b3e52b0753750642f463c5230af34ed4b0d12f5ef6e6" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/receipt.sha256", "bytes": 808, "sha256": "72ef9aa7ce5e9f65ce1e588962221f617779d2ef9434ea1e4801fa7563fab63d" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/resources.csv", "bytes": 6527, "sha256": "528feca547f2c2bd5f60aa28cd7eed2e4886fd98112297de68734092e5dca873" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/results.jsonl", "bytes": 1136, "sha256": "18bbe5a2addd0cf5b4069582a1bf7734b5912997241e2c76248ea3854459897a" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u128-pf0-20261004T223830Z-661175/runner-exit.txt", "bytes": 37, "sha256": "649b371d3b9514fe990433fdc386eb7e6a53a51b69b6f67de0b8003dad6cb26b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/bench.stderr.log", "bytes": 171382, "sha256": "d523809be9738a368cfd7bbe3c3cda3ed5d2d907c2ac70113cb0d54245ae7015" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/provenance.txt", "bytes": 2558, "sha256": "7fafec9133078a09d3b4c43fef6821fd41ab8a0f76f21b2b0c4abd7c41fa7c0d" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/receipt.sha256", "bytes": 808, "sha256": "37533aa0be0a7a8fe1741eee14f84bd132989a2b1f8c1d29d55d96778f6d5dca" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/resources.csv", "bytes": 4283, "sha256": "9dc2a2a1471c3f302402b8b69bbfddb7813fb4c641a52561a6e2c803be19cdcf" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/results.jsonl", "bytes": 1137, "sha256": "d987ae554284f51ae337a8240dc330b95614803937a9c4e2dfe042800b352d0e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u256-pf0-20261004T224144Z-665751/runner-exit.txt", "bytes": 37, "sha256": "ec14e66b2a5e96c3967f1f6a45b10c7fff8b2896225a09172eac94eeab620b8a" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/bench.stderr.log", "bytes": 171384, "sha256": "0401b0156c02b49438f4fe8bd6b88164917adf3a6317e2f2889629e235c24ab1" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/provenance.txt", "bytes": 2558, "sha256": "7b31b74c73c4830d3dd4308ecc2f828c8d2529acd0e6802c4a366481c46d975d" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/receipt.sha256", "bytes": 808, "sha256": "594032f4643a488797d4bbb2f3b5123d7eecfacb979811e4e0185e78386562b1" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/resources.csv", "bytes": 2855, "sha256": "2d5b9ccc729b501b70d7f3db8cddd5d875969caa616c8c1ec00b0a3a2567aa7f" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/results.jsonl", "bytes": 1137, "sha256": "0acc280b50cfbfbb574c9418f456eca8ffdb35be4929a62da218abc9811fd49c" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b2048-u512-pf0-20261004T224351Z-668259/runner-exit.txt", "bytes": 37, "sha256": "bc6f599c87139278bbdf1c5e9bdc1896ade63df13557e9d40108f49231fffd56" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/bench.stderr.log", "bytes": 171387, "sha256": "f2b56c7c084bbec2504655b5bc4c20673262c64818e97e8cc38243960b40060b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/provenance.txt", "bytes": 2560, "sha256": "50d2c7c3d472cc46ed00e3708febe6648a4248ab4779cdfd9d00ab92b3a24d94" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/receipt.sha256", "bytes": 812, "sha256": "07dbb0fb437dc5bfc85a3a923ee8b80e05327c2fbca0bace778a8fc67c5eb51d" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/resources.csv", "bytes": 1971, "sha256": "caab6db82ff2a0cff0c3625c94fab5dae78841c4046e2884ca54c96352dd0940" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/results.jsonl", "bytes": 1136, "sha256": "e65b653539096b7ab7b75b8a250ef9e323fbc7d74b98f9c48370c7efa9fea252" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u1024-pf0-20261004T225256Z-677442/runner-exit.txt", "bytes": 37, "sha256": "d276c55bffaa20f3292efd6f8640721638b569deb085d68a47696af353f737ad" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/bench.stderr.log", "bytes": 171382, "sha256": "6e44a3c5df66cdc402062662b12fab2e557797698a9f48c283ca1fa4b5b26d4b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/provenance.txt", "bytes": 2558, "sha256": "2fdbfe20ee2e6c7a6f89ff4c3775842acef90689bf9d5ce6337a2f0d363c5269" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/receipt.sha256", "bytes": 808, "sha256": "e0d89858c0df33421a6cca21b44fc2a47ba6629a4f2f23dee30a3cf4b3a6f31e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/resources.csv", "bytes": 6527, "sha256": "0eab2b3abdacf2369d59941ce8442dff4b9d43f683e5b84526f48750b4651c3e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/results.jsonl", "bytes": 1135, "sha256": "497cc81965d54ad11a2b30385c5556b0e5ce3c4ef2e5c49c95ba8d3d9f11ecbe" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u128-pf0-20261004T224611Z-670693/runner-exit.txt", "bytes": 37, "sha256": "29296f862c2795c63fdce62c88afe24b5e339d538d9b2874250089dc1912e94b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/bench.stderr.log", "bytes": 171382, "sha256": "88991cce1bff2ed0f6f1c843ce064cae38c364780377d5aec42e9837251fd8cd" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/provenance.txt", "bytes": 2558, "sha256": "3da7a5d4f4d8fd0a785e9dc784390077fed1874976d86337143957b86826eb91" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/receipt.sha256", "bytes": 808, "sha256": "0aaac7d327ea4ebb40088b243a7445009dd3f6eea5fbdd8104fd604f3046e06b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/resources.csv", "bytes": 4342, "sha256": "3ed1a34ce7dafaf05013ec63248383603778dedcfa1c698c9960a080c6ff2e3e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/results.jsonl", "bytes": 1136, "sha256": "c5c782ca8b6a7f64c8e6035acbc47dfccc1cb1c699487dc94aada1a0fa1655ea" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u256-pf0-20261004T224925Z-673871/runner-exit.txt", "bytes": 37, "sha256": "8556051deedd008ebfeca8472ab431489450febd7f4e3b790c35f02d28996c29" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/bench.stderr.log", "bytes": 171384, "sha256": "b4706767d3fce80952d695974decf2315bf23cca9d86066d4f91d5f8d2423ded" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/provenance.txt", "bytes": 2558, "sha256": "227772b1dfa903c8a503f7e0a76118d382a6c169ab9ae1fbffaacaa61647a4d2" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/receipt.sha256", "bytes": 808, "sha256": "b1c8dc28dac826123c33b6c960bbe7da6e180a0c732961ce65d92ac791386e57" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/resources.csv", "bytes": 2855, "sha256": "33556cb95bb5c67c1d9e94f5615a60b8abd9d6f994a6b58b7a15a7fac47e539e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/results.jsonl", "bytes": 1137, "sha256": "4173fb748d85397af9a100546c4ebf763b52de7bd280ca311421c02b81661e68" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b4096-u512-pf0-20261004T225132Z-676022/runner-exit.txt", "bytes": 37, "sha256": "e832373e8f5b7cbf77d64055274904e697590e115eb3e3ccd1a7f20daf113f93" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/bench.stderr.log", "bytes": 171380, "sha256": "f022d07972554fe0c81af278bf81b43008052c115469d71926f8584d2c4ecd65" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/provenance.txt", "bytes": 2556, "sha256": "97a389fef86bcf01e82875ade5af70877cf084acd9be0808aebb2ff2571324fa" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/receipt.sha256", "bytes": 804, "sha256": "588a3e6e428445a7095bfe7353ca081fbdc37a5add152bf0e16f5803726b753a" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/resources.csv", "bytes": 7225, "sha256": "86fb23f817995984e14455e7d4665401d0576e9f5006b8ce611f7ed712dd3368" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/results.jsonl", "bytes": 1135, "sha256": "24fa719439a54421148231b6365df83782ade33fa185293f0c3923d33062a191" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u128-pf0-20261004T222319Z-634105/runner-exit.txt", "bytes": 37, "sha256": "5eb9e05f35ab028bb3d3c3693eb7e35a7d9c2ec1b8c04c720f25a8aeffeee85c" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/bench.stderr.log", "bytes": 171380, "sha256": "2536b02c775e1b1d80ae631a8fccd4b703be70fdc50a40d496304cdff68e8ab2" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/provenance.txt", "bytes": 2556, "sha256": "d584ba4aaee934d05890bb082e13fde9c0ebd17f96d982bf668389199d3b489b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/receipt.sha256", "bytes": 804, "sha256": "969521c01ae7dd1bb8462475823b6911d36db8de1939c6d4fce176ed0eacfdc3" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/resources.csv", "bytes": 4351, "sha256": "c6aa6c0d8ae12ddbb03187d7c42552115c83f1e95d76f7b220ccd44b02d2255b" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/results.jsonl", "bytes": 1136, "sha256": "460a1232b5ca8ae8d0ccd0a61650911df8f8ab69ae7c1db75378682896b93786" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u256-pf0-20261004T222716Z-642697/runner-exit.txt", "bytes": 37, "sha256": "3c4a460c71d95d398215db2a9e2dcf9a26687dd1467db97991346d5442fe8e8e" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/bench.stderr.log", "bytes": 171382, "sha256": "09fead632bfc05d5195a950fac5461830c91b0fe3cda4121998d7c1bdbd0007c" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/provenance.txt", "bytes": 2556, "sha256": "52eae6326dfa74e90830f7061d34417f161256147c90e25b6a45592815e57185" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/receipt.sha256", "bytes": 804, "sha256": "c2b7d88c82e4f616b7ecd5a4fa9afd63584745dc41d6eba1fcc5ceaa0021be00" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/resources.csv", "bytes": 2855, "sha256": "7fbf19c96d4effe4378965d29cbb7269dde19fad543c24ab97abbb641b7b3ad9" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/results.jsonl", "bytes": 1136, "sha256": "5877447e02a26a5bf68db4e07da2b070635dfa0f4be4b64383c04e40fee43c7a" }, { "path": "historical180/benchmarks-20261004/prefill-c32-t14-a1-b512-u512-pf0-20261004T222924Z-646700/runner-exit.txt", "bytes": 37, "sha256": "282bc0afef262a0b5d8335e7adca2129ecc63125f063192bb1c05e8b6807aef4" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/bench.stderr.log", "bytes": 279211, "sha256": "e7a63280f08f7b2ec0f5599fa74b6ed20391af1f12fab12d6d475f3dd66f59b9" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/measurement-status.txt", "bytes": 36, "sha256": "2a06a51c62b3ac6caf7ca10bf48fbf89c7d81ea3ed79e6c0fb722cf2433104d5" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/provenance.txt", "bytes": 2154, "sha256": "a1a07a5796fcd6e50e178220c8f210b548dd0f9ed531e5ecebe2058bf9132219" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/receipt.sha256", "bytes": 796, "sha256": "d76e9d5b08b12a6d634cb9a8159e4dd97b962c1e4b99d0db903b9ea231d35b0a" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/resources.csv", "bytes": 11101, "sha256": "2f88900444e8d52dea0546c594b4dcfa93af9d415a992732126f25d2a651d78e" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/results.jsonl", "bytes": 13580, "sha256": "86c0646aec46a2797435ac3dc33ee3ef9bb0b4a0839a0dd8781fd4b4b9c72c77" }, { "path": "historical180/benchmarks-20261004/threads-c32-t4-6-8-12-14-16-a1-20261004T213654Z-567469/runner-exit.txt", "bytes": 37, "sha256": "468a8902a90d5d2deaa856df50fc7dc6061fdf9bb19823edd26fedda7f9e6afa" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/bench.stderr.log", "bytes": 171388, "sha256": "0014fde52ba5fc3cdc61599d3494ee8d1670868ea99b1dfc10d6bf0a614540fb" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/measurement-status.txt", "bytes": 48, "sha256": "c54d2fe9bb025ee57cf05f6ca1965b28498ee312ce83421f9195ddf53be1c996" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/provenance.txt", "bytes": 2652, "sha256": "76b7cff805c169c65d80bb8158651957579d595466c4c3c40a4057600527e837" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/receipt.sha256", "bytes": 1031, "sha256": "153351e63a2fa4dd0a7818cbe7a54f7ab04fae885cf15a5230f8da1da04d79d8" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/resources.csv", "bytes": 4204, "sha256": "a237a168941efb127deb451f53078c521305360cb94d24f863b50172999164a6" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/results.jsonl", "bytes": 1141, "sha256": "79b73c0a2b608355a47aa15a708ffe165066e9808e5aee0e332306c3dc2443bd" }, { "path": "historical180/benchmarks-pp8192-20261005/pp8192-c48-t14-a0-b8192-u8192-20261005T073938Z-1096850/runner-exit.txt", "bytes": 37, "sha256": "97193d97a3ff5d6003a778adf857368e148b5a0a98435f950f99b37219a6065b" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/bench.stderr.log", "bytes": 1168831, "sha256": "f68de80fedf7ffeba98e470eeb0e9fa787b8c6d918d634cfe0d8c781cea0c268" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/provenance.txt", "bytes": 2669, "sha256": "8de5eb652b9a96fb7ab771036360e232b4fcdd490c8cafcf74b011412a9d7ae1" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/receipt.sha256", "bytes": 1051, "sha256": "357247eb5a43ca30853f6ddd55c24b39a8d3f3cc31aa4c46282a0e4352a92181" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/resources.csv", "bytes": 6097, "sha256": "85d33afbdff77106c2ed7e7aef732e30de4a0aeb572b524fd573d678d06081a7" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/results.jsonl", "bytes": 2274, "sha256": "2b18017dda26cf600bd5a52c15ddbc61b72ccfe2e26fbb20e7f0a6cbea27556d" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b2048-u2048-nopo0-20261005T064446Z-1028912/runner-exit.txt", "bytes": 37, "sha256": "b835730cc8ab281595bb8a5ee5b8cc5c176086c8460ddc9761c51838427a7a28" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/bench.stderr.log", "bytes": 1168831, "sha256": "adbd453af27113f5f18f7c1e1b2657f3430765a5dc9991b6e265b89e4a15adf7" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/provenance.txt", "bytes": 2669, "sha256": "329446d3530a905f50e044c5d5d05ad36d470b71c5f1ed53270619fff03c894d" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/receipt.sha256", "bytes": 1051, "sha256": "e873f9c4472d530825e15fb561b26f1d9cae624d12521125309fcd25d9fc0fb2" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/resources.csv", "bytes": 4932, "sha256": "3eefcd4019b13eb2f7e941d1a049752d0414d1af135651220490fa6b0320e39c" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/results.jsonl", "bytes": 2270, "sha256": "75e37c24a1918c44ec3fa4f380427beca7fb1ea8013785f0046ecd977eb807c2" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u2048-nopo0-20261005T064617Z-1031775/runner-exit.txt", "bytes": 37, "sha256": "c2a31dcc3f5bea6733b6674c1de2971796a4afaff6c8a5bdff650ee4b935d0ea" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/bench.stderr.log", "bytes": 1168833, "sha256": "94d7ddcce02470afe7ddbd18fb17fc00908e6e62876c83e926dbb4184f43e3e3" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/provenance.txt", "bytes": 2669, "sha256": "152d2cc76bdebe4ca487e23e9298de41c3e1c325b32c611b0e5113b6df0c996a" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/receipt.sha256", "bytes": 1051, "sha256": "e5495efb33e1c24e73cacfa4916102361cdf217a25c16548f7c7dbae79808e13" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/resources.csv", "bytes": 4378, "sha256": "494b7d34beec36b9d0e9a8afb6801d121cc67cf24d715a2b1e486ca09fd8f538" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/results.jsonl", "bytes": 2270, "sha256": "254b2c28d7aa3ee4cf4d89817705c962b8d71704c6f5cb5922dad1cca717b117" }, { "path": "historical180/benchmarks-speed-20261005/speed-c48-t14-a0-b4096-u4096-nopo0-20261005T064730Z-1034279/runner-exit.txt", "bytes": 37, "sha256": "0913243fbcd759ed95713cdffc169df3ecdb1d6a5c7a0242dee35bcc5814467e" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/bench.stderr.log", "bytes": 1168831, "sha256": "00c718ec050f9a9016982f5c8b9b5deeda583385b517a113f18439962b968427" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/provenance.txt", "bytes": 2669, "sha256": "cd1b90794e92d64eac23a19147306fe9ba74ac99d97821ac64fed28538571086" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/receipt.sha256", "bytes": 1051, "sha256": "0b4f4e14eafcb0d5746fcf168dcd2af24eaf35f56b2c56ea09eaddc58df0ad1a" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/resources.csv", "bytes": 4899, "sha256": "49bf5ae9af36ea2c6600bed3435a5a19f34e6a9e52b63e500412d967415f0648" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/results.jsonl", "bytes": 2269, "sha256": "76dfec7805cae6e5518ce01791891a0a570db1e154fff436831d51c016651368" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b2048-u2048-nopo0-20261005T064835Z-1036566/runner-exit.txt", "bytes": 37, "sha256": "c336ba5336916504d1b27e557a5c092c5a5c81eb8046f2b7f1f74bfedd24b097" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/bench.stderr.log", "bytes": 1168831, "sha256": "20bbc976cc88e014dd557620355ad887f7a0dd8545c0d77a00891a80a0abba93" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/provenance.txt", "bytes": 2669, "sha256": "f0377fe7569f5ddd616ccaeb0d7fb78697d80a754dadb7b680d06d5d35fa447d" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/receipt.sha256", "bytes": 1051, "sha256": "818c4a5439be247da450fc19695b4b3178dcdd83d5cabf38d72bf5958d98c22d" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/resources.csv", "bytes": 4898, "sha256": "be49a5d8b15c0463cbfa9b10eeafb0f89ad28044958c66cb848de0978eef9fae" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/results.jsonl", "bytes": 2270, "sha256": "2025e9cd8dfff3b553dda790ef88aede231b4b2f30e6248540f9b7572f8c14f9" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u2048-nopo0-20261005T064948Z-1038971/runner-exit.txt", "bytes": 37, "sha256": "ef6288ac2aac10a385fe7a58441607102a584b21e8e760f460b6fb403ef078ea" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/bench.stderr.log", "bytes": 1168833, "sha256": "86311ab90559822dfb5871e6d5e112db1e8d311658c767c9241526985ed62682" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/provenance.txt", "bytes": 2669, "sha256": "7deecda7d7122c7e6ce7c260d8e9d4654d136dec0cf9cf797cc19cf220c75693" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/receipt.sha256", "bytes": 1051, "sha256": "33ea8cedfcfda8f36cf893ccc129c0f8333beb994f1605b118a7bf619d8f5e14" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/resources.csv", "bytes": 4278, "sha256": "ba2bf91b7010ca5e866736f2d0d674562d2601b71ea26482d3bc29e838f13069" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/results.jsonl", "bytes": 2268, "sha256": "830767544026a37f5e44b2b28c2a22a7b4d23df73fa35a8f7bb76cd4eecc9bd7" }, { "path": "historical180/benchmarks-speed-20261005/speed-c64-t14-a0-b4096-u4096-nopo0-20261005T065101Z-1041105/runner-exit.txt", "bytes": 37, "sha256": "6ccfb00d46cfd6817b19b22186b2402e2801a8509c15f44bc80e96cf4c39e680" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/bench.stderr.log", "bytes": 1168832, "sha256": "3d2115d3ad6a8a0c66b3cdd9baafcbe1110a7be63d2cd8fd6af01368da1f088f" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/provenance.txt", "bytes": 2669, "sha256": "10bc97868340c30498eddc10904bd017aa0176f4a8a51fcf679088bf43e2637c" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/receipt.sha256", "bytes": 1051, "sha256": "522f168f020d656ff973be7bef8b8e3d1e42a16b8fad8a2f5ef22b90675dc7b9" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/resources.csv", "bytes": 4828, "sha256": "02ac80cbba58f7a8723a556cf9d6959457a4c57604753b27f1337b5d4bab5807" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/results.jsonl", "bytes": 2270, "sha256": "db1d56c38bdd6a7dde6dc9df216375fd06a402f9c926d86ab2c3a858ece4a902" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b2048-u2048-nopo0-20261005T065205Z-1042879/runner-exit.txt", "bytes": 37, "sha256": "a17e76e8bacd20a7312e53a9021280723deb01162952d331d33c414b343944a7" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/bench.stderr.log", "bytes": 1168832, "sha256": "503d719ec6098cc2b77857855ca4fa785529620e7d21d105b0ea3649ffc804d3" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/provenance.txt", "bytes": 2669, "sha256": "11f777426880217c84a2b9e54efa850654c935af256d2d3b3f57845309cb57c8" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/receipt.sha256", "bytes": 1051, "sha256": "67d13d011620ff1c115fe4778316bb325b909d5dbf5d6b95fd0ba20919c7787e" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/resources.csv", "bytes": 4828, "sha256": "287b842e41f03a1a88b668d001b9221237819479d5203858171725fad966aa4c" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/results.jsonl", "bytes": 2270, "sha256": "ceac905722f6af1cd4d5b1858a2f7cdc8137aed25ef81e1e5d0d31620555d10e" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u2048-nopo0-20261005T065318Z-1044457/runner-exit.txt", "bytes": 37, "sha256": "24430e38e60d80b33f1d84f9e2f682edfbe6ba97a1688992bce54887a9fa9f34" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/bench.stderr.log", "bytes": 1168834, "sha256": "03494318c0e17eec4010a86627b766d0ec1dd05214800cf8b6b856efd7705a46" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/measurement-status.txt", "bytes": 48, "sha256": "8c1abed498610e88ebdc50a21a0108125138d8a46c9e0f0132dcf5c7b8571032" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/provenance.txt", "bytes": 2669, "sha256": "a0bbd27aba47bb47a2a617483fede87239d30d32cd2a6ddc477d10cfe45494da" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/receipt.sha256", "bytes": 1051, "sha256": "85e9c4123e8d8029ff2d1b58d98d0ab42b08fb74f21435e060b4855356134d57" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/resources.csv", "bytes": 4276, "sha256": "c27939a34bf070bbf8e9d9f20491dfb479fa121394446134c477dcb6a15b4015" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/results.jsonl", "bytes": 2269, "sha256": "c5c590ce00a30246e20ae85cf3b26d124faaebc74bdad7f22ac207a14aec41d7" }, { "path": "historical180/benchmarks-speed-20261005/speed-c80-t14-a0-b4096-u4096-nopo0-20261005T065430Z-1047277/runner-exit.txt", "bytes": 37, "sha256": "573bf31003b169602e164aacf19ea28bc95c37529ca194da9c968e99948d0a10" }, { "path": "profile-iq3-175-r1-20261005/chat.configuration.json", "bytes": 1314, "sha256": "0bfcdd54605c7874477e266e07d26c668d51fb060870b67f0bc3a2ba33bd4e11" }, { "path": "profile-iq3-175-r1-20261005/chat.csv", "bytes": 1892586, "sha256": "4714830e861748655f58281debdd1bb3bf2ea34312f1b1be9ee38ae547e5cd98" }, { "path": "profile-iq3-175-r1-20261005/chat.log", "bytes": 769, "sha256": "1375d76c0095457704f1fda42aa092c791b5a11c7b4bc8eca12528ffdffb440a" }, { "path": "profile-iq3-175-r1-20261005/chat.prompt.txt", "bytes": 1454, "sha256": "4c32a3ef9b38ddc0139371a96130118fb92f584e691ebbfd5ed9e9515fa3bf2e" }, { "path": "profile-iq3-175-r1-20261005/code.configuration.json", "bytes": 1314, "sha256": "dfcb2a4679ff720432d2cea8d88d70bf5119be1737066b787f022cb101bb16b8" }, { "path": "profile-iq3-175-r1-20261005/code.csv", "bytes": 1758300, "sha256": "6bae78a67fe5cfa4709f82443bdcf6fba6d66403a8a10057d84ab30454f84d3b" }, { "path": "profile-iq3-175-r1-20261005/code.log", "bytes": 769, "sha256": "035f9a42c0992a61be4d5cfd03e3caa4bf408dae46a5c987325401d3dfa3a278" }, { "path": "profile-iq3-175-r1-20261005/code.prompt.txt", "bytes": 1658, "sha256": "e30bf0e4894a7223bc6d0cabebf695764d77874bf814b12afd07773808850317" }, { "path": "profile-iq3-175-r1-20261005/resources.jsonl", "bytes": 168993, "sha256": "3afe220b178a7b0dfc1a472d844838d1d86fcb9d944dd15b22c620de1d838c0e" }, { "path": "profile-iq3-175-r1-20261005/routing-merged.csv", "bytes": 6736840, "sha256": "b6ea17b6a951bed511ad56bff039fe5e8288c0915c21365e8969d630ac604c83" }, { "path": "profile-iq3-175-r1-20261005/summary.json", "bytes": 9010, "sha256": "c336d60b426acb3ce73b9d26a841b945eb1851595ff147c73f1defb139d5e66f" }, { "path": "profile-iq3-175-r1-20261005/tools.configuration.json", "bytes": 1316, "sha256": "e2f3a79012392d2102af27dd3a6627a755d8774ac43962bac5df000716bbea73" }, { "path": "profile-iq3-175-r1-20261005/tools.csv", "bytes": 3085954, "sha256": "a5d16909bdfd42f64c4536012ded8b3a96d643fcba1374da414958e4059f0575" }, { "path": "profile-iq3-175-r1-20261005/tools.log", "bytes": 770, "sha256": "eedbdc2d728f8ff8733a1e9b8f029a15ee9516f6bb322c4f45ad91c1dec2b95d" }, { "path": "profile-iq3-175-r1-20261005/tools.prompt.txt", "bytes": 4445, "sha256": "0175cebe840ffb49083b4c0111c8a228386d3683fd0be3bc29950e50f77af5d0" }, { "path": "historical180/profile-iq3-175-20261005/code.configuration.json", "bytes": 1339, "sha256": "cd00480ee4e11c8c6b73511fde9fb2ae43c67c9036d23a2733ef1042118cbe8a" }, { "path": "historical180/profile-iq3-175-20261005/code.log", "bytes": 186, "sha256": "c62ef0182be03707001252a16ae7ae37f6519f403f8d1374a3b75d06f6813cc8" }, { "path": "historical180/profile-iq3-175-20261005/code.prompt.txt", "bytes": 1658, "sha256": "e30bf0e4894a7223bc6d0cabebf695764d77874bf814b12afd07773808850317" }, { "path": "historical180/profile-iq3-175-20261005/resources.jsonl", "bytes": 58504, "sha256": "0b6ae2bf7d1f00985aae47f8018091352959701ca203cf41335a3bd8602b3321" }, { "path": "historical180/profile-iq3-175-20261005/summary.json", "bytes": 8227, "sha256": "d1ed74111150099b799821e7ebcfdf5dd3a820f91ad47752023d3e4b56b47c5f" }, { "path": "historical180/profiles-20261004/chat.csv", "bytes": 1893213, "sha256": "dd23cc0c8787b51ebff9fcfdc6c865f9f879ae3ec8a13f9e8ce2da508a2d4abc" }, { "path": "historical180/profiles-20261004/chat.plain-tokenize.request.json", "bytes": 1527, "sha256": "12b9af3009b67d22866b615eb57ac98650c79c135374e870f085c85990353b00" }, { "path": "historical180/profiles-20261004/chat.plain-tokens.json", "bytes": 2056, "sha256": "f28725535a9d10c5e7d4d070c58851652da0689779cbfb11a26b9bd46fae9165" }, { "path": "historical180/profiles-20261004/chat.prompt.txt", "bytes": 1454, "sha256": "4c32a3ef9b38ddc0139371a96130118fb92f584e691ebbfd5ed9e9515fa3bf2e" }, { "path": "historical180/profiles-20261004/chat.server-template.txt", "bytes": 1722, "sha256": "c37c272d6b77bf7bb6caad78a739a0895fa0bd1c5c0edf659b546a168b3843c1" }, { "path": "historical180/profiles-20261004/chat.template-tokenize.request.json", "bytes": 1798, "sha256": "5b4f9ad13e26aae98719e35da3e64aea262df3124cf47edd27961a3faaa3dcb7" }, { "path": "historical180/profiles-20261004/chat.template-tokens.json", "bytes": 2279, "sha256": "8e155e2062ace1bead149738f209a93e3cecda0beca442a1cdf36643f8e7c22b" }, { "path": "historical180/profiles-20261004/chat.template.json", "bytes": 1743, "sha256": "70c8c9c59566334bfcd4366214dbbdc96fd0e302cb3336ad825890409ccdb141" }, { "path": "historical180/profiles-20261004/chat.trace.log", "bytes": 758, "sha256": "df7552dc8978b6be7427e2e2f5289e77150a8aa98e80121f71b3a0df67cefb3f" }, { "path": "historical180/profiles-20261004/chat.validation.txt", "bytes": 70, "sha256": "103aa681d2242e6ee9b8f29d43b408fa85715ad5e2ae34ec45efe22751b55bfd" }, { "path": "historical180/profiles-20261004/code.csv", "bytes": 1758279, "sha256": "1af1b3de68160ed7d21b706ee173ef6c8879b9dc7577513a17cdc2b9ef3e5d99" }, { "path": "historical180/profiles-20261004/code.plain-tokenize.request.json", "bytes": 1731, "sha256": "e68dc65874f9e384e79a01ddc54037c8a684cee7cfead2f9958f3ee696441d9c" }, { "path": "historical180/profiles-20261004/code.plain-tokens.json", "bytes": 1491, "sha256": "f9095279f74b158bde0559aba625b084cf4e09ff452776ba4876fcb9413e9265" }, { "path": "historical180/profiles-20261004/code.prompt.txt", "bytes": 1658, "sha256": "e30bf0e4894a7223bc6d0cabebf695764d77874bf814b12afd07773808850317" }, { "path": "historical180/profiles-20261004/code.server-template.txt", "bytes": 1926, "sha256": "d3dd1a7c0e482078f1907dd171da662e307bfa57727733577e13d2726f6e87e4" }, { "path": "historical180/profiles-20261004/code.template-tokenize.request.json", "bytes": 2002, "sha256": "349548b9c26ee37bf7309648404260b63dd22877f426ac53d908c5bf73e145ef" }, { "path": "historical180/profiles-20261004/code.template-tokens.json", "bytes": 1712, "sha256": "31bbfff5523cc8d47f6e8070ea580c6ebde6c1b9dc920f2a094b9369b07ba04d" }, { "path": "historical180/profiles-20261004/code.template.json", "bytes": 1947, "sha256": "e1629acb156c2a37181c925db077160a995c95030bfb59833abea4b833fe84dd" }, { "path": "historical180/profiles-20261004/code.trace.log", "bytes": 758, "sha256": "0aebd5731a5ddffa09a5ea4a1e32fbd0a1e4e057a8a9714d717ed7635df4986d" }, { "path": "historical180/profiles-20261004/code.validation.txt", "bytes": 70, "sha256": "5840d5911e5256e842c12d21f58a3d61bb2326632cc075aa7449b03e2c4e794e" }, { "path": "historical180/profiles-20261004/profile-artifacts.sha256", "bytes": 1126, "sha256": "bf2f178d663b070b829a3c42b85cae604cfb91b7a47b5e3ce94f589764c33056" }, { "path": "historical180/profiles-20261004/profile-complete.txt", "bytes": 51, "sha256": "f60716adb6aa071fc8dda2b259e19549964c22b0d99a457b54248e148cdc6175" }, { "path": "historical180/profiles-20261004/profile-limitations.txt", "bytes": 161, "sha256": "36012fe41e3dd7018a1cbcd4a3c2895a255f1773b682a4525980bbe9a90e044e" }, { "path": "historical180/profiles-20261004/profile-provenance.json", "bytes": 3878, "sha256": "dc2882a4a45ff8a28db37cabcdfa5db112ee9179478f4cb5dcdbc9ef811b82de" }, { "path": "historical180/profiles-20261004/resource-guard.log", "bytes": 0, "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" }, { "path": "historical180/profiles-20261004/resources.csv", "bytes": 5764, "sha256": "4a2dbaf6e0b3e7af67a483f10e049a81cb1ace8c902bb9c95e7969850b33c36f" }, { "path": "historical180/profiles-20261004/routing-merged.csv", "bytes": 6730829, "sha256": "a31bde43aafc447b2b7df8aa02c0cd281e05205a9d73cc06ea2669964237f269" }, { "path": "historical180/profiles-20261004/template-health.json", "bytes": 15, "sha256": "a29ee2b15c494311c52521766e44af56a3ad2248e7a8ab465e5206463c13d288" }, { "path": "historical180/profiles-20261004/template-server.log", "bytes": 1291, "sha256": "98b8fed231f6bcd35f7d7a60b8dbee9961a7df8b6e8629f0422879cf5f33a00b" }, { "path": "historical180/profiles-20261004/tools.csv", "bytes": 3079337, "sha256": "e3daa60d63635c40d9ee4073cebd549c30d6d9f3974e60cfb151763745036109" }, { "path": "historical180/profiles-20261004/tools.plain-tokenize.request.json", "bytes": 4740, "sha256": "351adb2959be80255d71e6fdff415cd6eaebcc89144d5710d497aaa3122ea147" }, { "path": "historical180/profiles-20261004/tools.plain-tokens.json", "bytes": 4536, "sha256": "43945d30bd1e32b8b63d368df8efc224c403ecf348b03716fff587e34a12d92e" }, { "path": "historical180/profiles-20261004/tools.prompt.txt", "bytes": 4445, "sha256": "0175cebe840ffb49083b4c0111c8a228386d3683fd0be3bc29950e50f77af5d0" }, { "path": "historical180/profiles-20261004/tools.server-template.txt", "bytes": 5758, "sha256": "ee938ab133247011e8bbf77b4614f780069d44375095f41e65a57ffd5dade17d" }, { "path": "historical180/profiles-20261004/tools.template-tokenize.request.json", "bytes": 6060, "sha256": "480a94d190108bed475851a42089874dc8153fbc7406535f9a53cc6ce95926b1" }, { "path": "historical180/profiles-20261004/tools.template-tokens.json", "bytes": 5839, "sha256": "8cf4931f56aedc59b304ce39cfccef13713c97097f253619a4fd5919fb621e48" }, { "path": "historical180/profiles-20261004/tools.template.json", "bytes": 6005, "sha256": "81acf65680c5fb21b43f598a8ee47f953233213bbf4e8229302f019ef83e154a" }, { "path": "historical180/profiles-20261004/tools.trace.log", "bytes": 759, "sha256": "b670c4c23070ea20e6c7cfa91de3128febf995a530189fe8c6fb62ac2f60633a" }, { "path": "historical180/profiles-20261004/tools.validation.txt", "bytes": 71, "sha256": "c5006179d8d963a7f0fb68528befc2b8c5ddfbd8a34e4c63f1ad4f0b9ac73110" } ], "privacy": { "policy": "Allowlisted synthetic metadata only. No transcripts, prompts, audio, credentials, personal endpoint or device UUID.", "path_normalization": "/filename replaces machine-specific absolute locations; numerical flags/env values retained.", "raw_files_published": false, "hash_meaning": "SHA256 identifies the original private source bytes; it is not a checksum of the redacted export." }, "totals": { "runs": 85, "measurements": 406, "configurations": 96, "measurement_classes": { "synthetic-speed": 240, "context-quality": 103, "native-random": 56, "warmup": 2, "functional-transport": 5 }, "native_samples": 168, "source_manifest_entries": 2217 }, "build_recipes": [ { "engine": "llama.cpp upstream", "source": { "path": "jarvis-flash174/upstream178-build.sh", "bytes": 2186, "sha256": "b46624daa26e715a97f149d9aaf39d700d2a794d342f242b157595a0ae1d2eba" }, "recorded_cmake_command": "cmake -S \"$stage/source\" -B \"$stage/build-cuda86\" -G Ninja \\\n -DCMAKE_BUILD_TYPE=Release -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86 \\\n -DGGML_NATIVE=ON -DGGML_BACKEND_DL=OFF -DLLAMA_BUILD_TESTS=OFF \\\n -DLLAMA_BUILD_EXAMPLES=ON -DLLAMA_BUILD_TOOLS=ON -DLLAMA_BUILD_SERVER=ON \\\n -DLLAMA_BUILD_UI=OFF -DLLAMA_USE_PREBUILT_UI=OFF \\\n -DCMAKE_INSTALL_RPATH='$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON \\\n -DCMAKE_EXE_LINKER_FLAGS='-Wl,-rpath-link,/usr/local/cuda/lib64/stubs'", "scope": "Saved compiler recipe; not a new build. Runtime flags are separate." }, { "engine": "ik_llama.cpp", "source": { "path": "jarvis-flash174/ik178-build.sh", "bytes": 2793, "sha256": "28656824d4842236b9924952fd2efda768269114c93bd0b6e1940031d6fd6f4a" }, "recorded_cmake_command": "cmake -S \"$stage/source\" -B \"$stage/build-cuda86\" -G Ninja \\\n -DCMAKE_BUILD_TYPE=Release -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86 \\\n -DGGML_NATIVE=ON -DGGML_AVX512=ON -DGGML_AVX512_VBMI=ON \\\n -DGGML_AVX512_VNNI=ON -DGGML_AVX512_BF16=ON -DGGML_SCHED_MAX_COPIES=1 \\\n -DLLAMA_BUILD_TESTS=ON -DLLAMA_BUILD_EXAMPLES=ON -DLLAMA_BUILD_SERVER=ON -DGGML_NCCL=OFF \\\n -DLLAMA_CURL=OFF -DCMAKE_EXPORT_COMPILE_COMMANDS=ON \\\n -DCMAKE_INSTALL_RPATH='$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON \\\n -DCMAKE_EXE_LINKER_FLAGS='-Wl,-rpath-link,/usr/local/cuda/lib64/stubs'", "scope": "Saved compiler recipe; not a new build. Runtime flags are separate." }, { "engine": "OptLlama", "source": { "path": "jarvis-flash174/opt181-build.sh", "bytes": 2109, "sha256": "71dfb27ad7df6d0f2f08ec6ee6c70b6cad4faaddb273238edb0d6dfcf3378b9d" }, "recorded_cmake_command": "cmake -S \"$stage/source\" -B \"$stage/build-cuda86\" -G Ninja \\\n -DCMAKE_BUILD_TYPE=Release -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86 \\\n -DGGML_NATIVE=ON -DGGML_BACKEND_DL=OFF -DGGML_BLAS=OFF -DGGML_NCCL=OFF \\\n -DLLAMA_BUILD_TESTS=ON -DLLAMA_BUILD_EXAMPLES=ON -DLLAMA_BUILD_TOOLS=ON \\\n -DLLAMA_BUILD_SERVER=ON -DLLAMA_BUILD_APP=OFF \\\n -DLLAMA_BUILD_UI=OFF -DLLAMA_USE_PREBUILT_UI=OFF \\\n -DCMAKE_EXPORT_COMPILE_COMMANDS=ON \\\n -DCMAKE_INSTALL_RPATH='$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON \\\n -DCMAKE_EXE_LINKER_FLAGS='-Wl,-rpath-link,/usr/local/cuda/lib64/stubs'", "scope": "Saved compiler recipe; not a new build. Runtime flags are separate." }, { "engine": "Strata", "source": { "path": "jarvis-flash174/strata188/build.sh", "bytes": 727, "sha256": "62760e503314258134a5f84b010f605d8bba723a8465964fc505d2f5061c175b" }, "recorded_cmake_command": "cmake -S /source -B /output/build-cuda86 -G Ninja \\\n -DCMAKE_BUILD_TYPE=Release -DSTRATA_ENABLE_CUDA=ON \\\n -DSTRATA_BUILD_TESTS=OFF -DCMAKE_CUDA_ARCHITECTURES=86 \\\n -DCMAKE_CUDA_COMPILER=/usr/local/cuda/bin/nvcc \\\n -DSTRATA_GGML_DIR=/source/third_party/llama.cpp", "scope": "Saved compiler recipe; not a new build. Runtime flags are separate." } ] }