{
  "humaneval": {
    "label": "HumanEval",
    "total_tasks": 164,
    "passed_count": 75,
    "success_pct": 45.73,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_humaneval_full.json",
    "report_sha256": "6d90478839d515483391ebbd198bf1832d853758f69e726ee01e89bda958b01c",
    "timestamp": 1789915634
  },
  "tooluse": {
    "label": "Telos tool-use",
    "total_tasks": 30,
    "passed_count": 19,
    "success_pct": 63.33,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_tooluse_full.json",
    "report_sha256": "5152eb231db643a4a13edda57f9bbc37cc4365696806d58717b1a1382b28aa9c",
    "timestamp": 1789918849
  },
  "cyber": {
    "label": "Telos cybersecurity",
    "total_tasks": 50,
    "passed_count": 12,
    "success_pct": 24.0,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_cyber_full.json",
    "report_sha256": "8db7a507c5a8c267ebfd6dca493220de1f9a6497c148502920c75d0d922985c8",
    "timestamp": 1789918896
  },
  "gpqa_diamond": {
    "label": "GPQA Diamond",
    "total_tasks": 198,
    "passed_count": 41,
    "success_pct": 20.71,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_gpqa_diamond_full.json",
    "report_sha256": "a840c250b513cc6e2b7792ec1bb0c7bfc90d750077f2bd101387d38376fb594a",
    "timestamp": 1789919328
  },
  "mmlu_science": {
    "label": "MMLU Science selection",
    "total_tasks": 833,
    "passed_count": 609,
    "success_pct": 73.11,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_mmlu_science_full.json",
    "report_sha256": "924866dd3067421133128336cc566141555530885e674bd9032eb7716447d382",
    "timestamp": 1789920919
  },
  "competition_math": {
    "label": "Competition MATH selection",
    "total_tasks": 700,
    "passed_count": 305,
    "success_pct": 43.57,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_competition_math_full.json",
    "report_sha256": "2aaf3bec80b13d769ef50d2167cd6230b5ffcd460b9870bf0dc06d1c05565c3f",
    "timestamp": 1789929373
  },
  "arc": {
    "label": "ARC-Challenge",
    "total_tasks": 1172,
    "passed_count": 698,
    "success_pct": 59.56,
    "retry_flags": {},
    "report_file": "logs/eval_report_lfm8b_arc_full.json",
    "report_sha256": "d748e4901d156d1ff58738b0b347ad78c9ccd3534ae092f60429c10dbe0a01a1",
    "timestamp": 1789933601
  },
  "_provenance": {
    "audited": "2026-10-28",
    "metric": "saved success rate; not controlled pass@1",
    "model_name": "LiquidAI LFM 2.5 8B A1B",
    "summary_discrepancies": []
  }
}
