{
  "schema_version": 1,
  "timing_scope": "model load through evaluator process exit; setup and transfer excluded",
  "clean_unresumed_run": true,
  "hostname": "9918f2a39d4e",
  "gpu_count": 1,
  "gpu": "NVIDIA H100 80GB HBM3, 81559 MiB, GPU-905b7c32-37d4-4a22-c33e-1bb5e6636d32, 580.126.09, 700.00 W",
  "python": "Python 3.11.10",
  "started_utc": "2026-08-10T15:49:18.154368+00:00",
  "ended_utc": "2026-08-10T20:38:59.191782+00:00",
  "wall_seconds": 17381.037394484505,
  "aggregate_h100_hours": 4.828065942912363,
  "task_count": 400,
  "h100_hours_per_task": 0.012070164857280907,
  "arc_trm_convention_usd_per_h100_hour": 1.0,
  "arc_trm_convention_total_usd": 4.828065942912363,
  "arc_trm_convention_usd_per_task": 0.012070164857280907,
  "returncode": 0,
  "command": [
    "/root/venvs/arc-ttt-h100/bin/python",
    "spark-gpt/evaluate_arc_ttt_aggressive.py",
    "--model-path",
    "spark-gpt/checkpoints/arc_rearc_50m_3kpertask_seed0_20260807/hf_step0030516",
    "--data-dir",
    "data/arc-agi-evaluation",
    "--output-dir",
    "evaluations/ttt_50m_3b_h100_sxm_seed57_full400_20260810",
    "--seed",
    "57",
    "--max-examples",
    "250",
    "--ttt-sampler",
    "legacy",
    "--max-seq-length",
    "16384",
    "--epochs",
    "2",
    "--batch-size",
    "2",
    "--muon-lr",
    "0.004",
    "--adamw-lr",
    "0.0005",
    "--muon-momentum",
    "0.95",
    "--weight-decay",
    "0.01",
    "--warmup-steps",
    "10",
    "--lr-decay-factor",
    "0.1",
    "--max-new-tokens",
    "1024",
    "--generation-batch-size",
    "16",
    "--attn-implementation",
    "flash_attention_2",
    "--progress-every",
    "1",
    "--no-resume"
  ],
  "summary_path": "/root/arc/evaluations/ttt_50m_3b_h100_sxm_seed57_full400_20260810/summary.json"
}