{
  "status": "passed",
  "accuracy_evaluated": false,
  "source_code_mounted": false,
  "device": "cpu",
  "requested_cpu_cores": 4,
  "requested_memory_mib": 8192,
  "cache_size": 0,
  "warmups_per_workload": 3,
  "measurements": [
    {
      "name": "short-2",
      "candidate_count": 2,
      "target_joint_tokens": 32,
      "actual_joint_token_counts": [
        30,
        30
      ],
      "state_tokens": 24,
      "state_sha256": "80584745717c23f0a3ea5b48a333550266fb3616a2f2f7e83f374410eae90b8f",
      "warmups": 3,
      "repeats": 10,
      "samples_s": [
        0.1230225140000023,
        0.11580429099999989,
        0.13012134300000255,
        0.13055259400000097,
        0.13944391500000108,
        0.1237515120000019,
        0.1800441300000024,
        0.12156313499999882,
        0.1251881130000001,
        0.11749492499999903
      ],
      "p50_ms": 124.469812500001,
      "p95_ms": 161.7740332500018,
      "min_ms": 115.80429099999989,
      "max_ms": 180.0441300000024,
      "sequential_requests_per_s": 7.651188603886277,
      "candidate_pairs_per_s": 15.302377207772555,
      "process_peak_rss_mib": 2599.75390625
    },
    {
      "name": "short-4",
      "candidate_count": 4,
      "target_joint_tokens": 32,
      "actual_joint_token_counts": [
        30,
        30,
        30,
        30
      ],
      "state_tokens": 24,
      "state_sha256": "80584745717c23f0a3ea5b48a333550266fb3616a2f2f7e83f374410eae90b8f",
      "warmups": 3,
      "repeats": 10,
      "samples_s": [
        0.1974090710000027,
        0.18758842100000095,
        0.1852979459999986,
        0.20146656800000073,
        0.2277730759999983,
        0.18973696799999828,
        0.18919480899999996,
        0.18587308600000085,
        0.18787166199999916,
        0.18382430499999813
      ],
      "p50_ms": 188.53323549999956,
      "p95_ms": 215.93514739999935,
      "min_ms": 183.82430499999813,
      "max_ms": 227.7730759999983,
      "sequential_requests_per_s": 5.165193443994345,
      "candidate_pairs_per_s": 20.66077377597738,
      "process_peak_rss_mib": 2599.75390625
    },
    {
      "name": "medium-2",
      "candidate_count": 2,
      "target_joint_tokens": 128,
      "actual_joint_token_counts": [
        128,
        128
      ],
      "state_tokens": 122,
      "state_sha256": "1092ffd94794efd83606ad4197d402fd31de3b64e1b6913b7575930a33ed4235",
      "warmups": 3,
      "repeats": 10,
      "samples_s": [
        0.38965914399999946,
        0.36982004799999757,
        0.4219474740000031,
        0.39345802500000104,
        0.4372249950000011,
        0.4004029459999998,
        0.4065815750000006,
        0.38176413199999715,
        0.37625566200000193,
        0.40223263900000106
      ],
      "p50_ms": 396.93048550000043,
      "p95_ms": 430.350110550002,
      "min_ms": 369.82004799999754,
      "max_ms": 437.2249950000011,
      "sequential_requests_per_s": 2.5129753461236524,
      "candidate_pairs_per_s": 5.025950692247305,
      "process_peak_rss_mib": 2599.75390625
    },
    {
      "name": "medium-4",
      "candidate_count": 4,
      "target_joint_tokens": 128,
      "actual_joint_token_counts": [
        128,
        128,
        128,
        128
      ],
      "state_tokens": 122,
      "state_sha256": "1092ffd94794efd83606ad4197d402fd31de3b64e1b6913b7575930a33ed4235",
      "warmups": 3,
      "repeats": 10,
      "samples_s": [
        0.7252715499999987,
        0.7080788560000002,
        0.740636140000003,
        0.7027773290000034,
        0.6812326490000018,
        0.6923296439999973,
        0.6939130290000008,
        0.6656062389999988,
        0.6926343539999991,
        0.710786358
      ],
      "p50_ms": 698.3451790000022,
      "p95_ms": 733.7220745000011,
      "min_ms": 665.6062389999988,
      "max_ms": 740.636140000003,
      "sequential_requests_per_s": 1.425869172646718,
      "candidate_pairs_per_s": 5.703476690586872,
      "process_peak_rss_mib": 2599.75390625
    },
    {
      "name": "long-2",
      "candidate_count": 2,
      "target_joint_tokens": 512,
      "actual_joint_token_counts": [
        512,
        512
      ],
      "state_tokens": 506,
      "state_sha256": "20b20e06825944d8a282379c72fa92e0d86b83167a7f1cc11ba569812f3d4552",
      "warmups": 3,
      "repeats": 5,
      "samples_s": [
        1.3366263479999958,
        1.4227291730000005,
        1.3784461649999997,
        1.3648940570000008,
        1.3449409340000003
      ],
      "p50_ms": 1364.8940570000009,
      "p95_ms": 1413.8725714000002,
      "min_ms": 1336.6263479999957,
      "max_ms": 1422.7291730000004,
      "sequential_requests_per_s": 0.7301789268104889,
      "candidate_pairs_per_s": 1.4603578536209778,
      "process_peak_rss_mib": 2599.75390625
    }
  ],
  "scope": "Sequential in-process top-level rank calls, including validation, tokenization, encoder, scalar head and response construction; no HTTP or network in measured warm calls.",
  "percentile_note": "Interpolated p50/p95 from ten samples (five for optional long input); not production tail latency or an accuracy benchmark.",
  "installation": {
    "package_version": "0.1.0",
    "distribution_source": "PyPI wheel, installed by pip; no editable/local source",
    "module_sha256": "66fccdd0df22c991447958debd94d4f07ce3723ca4add461c0ca12ed42c2bd45",
    "pypi_artifact": {
      "url": "https://files.pythonhosted.org/packages/12/9f/359c665e94b65ff9d7c3c69f6c9761d02f8de59497c10a4831748af6087b/gemmadecision-0.1.0-py3-none-any.whl",
      "archive_info": {
        "hash": "sha256=5897050c6e54deca3a5a683fa8a7a30ae2e2d1b496ae60e3362edca17d3cef52",
        "hashes": {
          "sha256": "5897050c6e54deca3a5a683fa8a7a30ae2e2d1b496ae60e3362edca17d3cef52"
        }
      }
    }
  },
  "environment": {
    "python": "3.11.12",
    "platform": "Linux-4.19.0-gvisor-x86_64-with-glibc2.36",
    "cpu_model": "unknown",
    "cpu_affinity_count": 4,
    "torch_threads": 4,
    "torch_interop_threads": 1,
    "versions": {
      "torch": "2.14.0+cpu",
      "transformers": "5.17.0",
      "safetensors": "0.8.0",
      "tokenizers": "0.23.2",
      "pydantic": "2.13.5"
    }
  },
  "imports_and_distribution_checks_s": 3.8222307370000004,
  "download_s": 2.9532738169999995,
  "public_model": {
    "repo": "rajan2k/GemmaDecision-270M",
    "revision": "785d530221c990671f29976902540101bb9c7647",
    "authentication": "Anonymous: package downloader uses token=False",
    "cache": "Fresh temporary HF cache; no model in image or mounted volume"
  },
  "load_and_integrity_checks_s": 9.205115679999999,
  "engine": {
    "model": "rajan2k/GemmaDecision-270M@785d530221c990671f29976902540101bb9c7647",
    "backend": "torch",
    "cache_size": 0,
    "temperature": 4.136820402388508,
    "max_state_tokens": 2048,
    "max_action_tokens": 768,
    "device": "cpu",
    "dtype": "float32",
    "head_dtype": "float32",
    "strict": false,
    "max_batch_tokens": 8192,
    "max_batch_size": 32,
    "model_revision": "785d530221c990671f29976902540101bb9c7647",
    "verified": true
  },
  "first_rank_after_load_s": 0.12441567300000145,
  "cold_download_load_first_rank_s": 12.28280517,
  "process_peak_rss_mib": 2599.75390625,
  "measurement_process_seconds": 45.701928091,
  "function_seconds": 46.516934946,
  "estimated_function_usd": 0.0032636281558113597,
  "rate_usd_per_second": 7.015999999999999e-05,
  "estimate_scope": "Function reserved CPU/RAM only; excludes image build/startup/storage/transfer",
  "harness_sha256": "84107693c60e55c8a5701bd1e47927069258c1fd050340e4b51a29736f61d656"
}