{
  "adapter_checkpoint": "/lustre/fs1/home/sa305415/shohin/artifacts/product_reasoning/qwen3.5-9b-c202236/b1_u256_dddb36f/checkpoint_0000256.pt",
  "adapter_checkpoint_sha256": "854a7cc44fbc2b54418f4e5bd09b7efeed0da44fc9ce217b0bb6b1997b722971",
  "adapter_metadata": {
    "arm": "baseline",
    "backbone_layout": "multimodal-language-model",
    "data": "/lustre/fs1/home/sa305415/shohin/artifacts/product_reasoning/data/v10_tokenbalanced_35m20c10s10p25t_4m_verified_r1.jsonl",
    "data_seed": 20260802,
    "data_sha256": "2461d6f70b44a142854d56c24e1fb42d600065e5788a2c4e055ba47b12696549",
    "loaded_model_root": "/tmp/sa305415/745345/model",
    "lora_alpha": 16.0,
    "lora_layers": 4,
    "lora_projection_count": 31,
    "lora_rank": 8,
    "model_loader": "multimodal",
    "model_revision": "c202236235762e1c871ad0ccb60c8ee5ba337b9a",
    "model_root": "/lustre/fs1/home/sa305415/shohin/artifacts/external/qwen3.5-9b-c202236",
    "seed": 2026080711,
    "selected_rows": 26387,
    "trainable_parameters": 2704896,
    "unfreeze_layers": 0,
    "update": 256,
    "warm_start_checkpoint": null,
    "warm_start_data_sha256": null,
    "warm_start_sha256": null,
    "warm_start_update": null,
    "workspace_architecture_sha256": null,
    "workspace_config": null
  },
  "arm": "independent",
  "backbone_learning_rate": 2e-06,
  "checkpoint": "/lustre/fs1/home/sa305415/shohin/artifacts/product_reasoning/aqc1_38bc1f3_r1/independent/commit.pt",
  "checkpoint_sha256": "185574eb495e5f5a8c5f1ad85b5e39dc5a5aeaa4a5ee8a21beb5f859131c3e52",
  "development": {
    "bbh_logic": {
      "control_accuracy": 0.3333333333333333,
      "control_commit_rate": 0.6239316239316239,
      "control_commits": 73,
      "control_correct": 39,
      "disagreement_selected_correct": 15,
      "disagreement_selection_accuracy": 0.8823529411764706,
      "disagreements": 17,
      "idr1_accuracy": 0.4444444444444444,
      "idr1_correct": 52,
      "oracle_accuracy": 0.46153846153846156,
      "oracle_correct": 54,
      "order_consistency": 1.0,
      "order_consistent": 117,
      "selected_accuracy": 0.4444444444444444,
      "selected_correct": 52,
      "total": 117
    },
    "math500": {
      "control_accuracy": 0.2619047619047619,
      "control_commit_rate": 0.35714285714285715,
      "control_commits": 45,
      "control_correct": 33,
      "disagreement_selected_correct": 27,
      "disagreement_selection_accuracy": 0.9,
      "disagreements": 30,
      "idr1_accuracy": 0.35714285714285715,
      "idr1_correct": 45,
      "oracle_accuracy": 0.42857142857142855,
      "oracle_correct": 54,
      "order_consistency": 1.0,
      "order_consistent": 126,
      "selected_accuracy": 0.40476190476190477,
      "selected_correct": 51,
      "total": 126
    },
    "mbpp": {
      "control_accuracy": 0.8,
      "control_commit_rate": 0.0,
      "control_commits": 0,
      "control_correct": 4,
      "disagreement_selected_correct": 0,
      "disagreement_selection_accuracy": null,
      "disagreements": 0,
      "idr1_accuracy": 0.8,
      "idr1_correct": 4,
      "oracle_accuracy": 0.8,
      "oracle_correct": 4,
      "order_consistency": 1.0,
      "order_consistent": 5,
      "selected_accuracy": 0.8,
      "selected_correct": 4,
      "total": 5
    },
    "overall": {
      "control_accuracy": 0.3064516129032258,
      "control_commit_rate": 0.47580645161290325,
      "control_commits": 118,
      "control_correct": 76,
      "disagreement_selected_correct": 42,
      "disagreement_selection_accuracy": 0.8936170212765957,
      "disagreements": 47,
      "idr1_accuracy": 0.40725806451612906,
      "idr1_correct": 101,
      "oracle_accuracy": 0.45161290322580644,
      "oracle_correct": 112,
      "order_consistency": 1.0,
      "order_consistent": 248,
      "selected_accuracy": 0.4314516129032258,
      "selected_correct": 107,
      "total": 248
    }
  },
  "development_maximum_swap_error": 0.0,
  "development_prompt_truncated": 0,
  "elapsed_seconds": 1160.554763623979,
  "gradient_accumulation": 8,
  "head_learning_rate": 0.0002,
  "head_parameters": 2237441,
  "head_width": 512,
  "holdout": {
    "bbh_logic": {
      "control_accuracy": 0.488,
      "control_commit_rate": 0.6384,
      "control_commits": 399,
      "control_correct": 305,
      "disagreement_selected_correct": 63,
      "disagreement_selection_accuracy": 0.8513513513513513,
      "disagreements": 74,
      "idr1_accuracy": 0.5616,
      "idr1_correct": 351,
      "oracle_accuracy": 0.584,
      "oracle_correct": 365,
      "order_consistency": 1.0,
      "order_consistent": 625,
      "selected_accuracy": 0.5664,
      "selected_correct": 354,
      "total": 625
    },
    "math500": {
      "control_accuracy": 0.26570048309178745,
      "control_commit_rate": 0.3397745571658615,
      "control_commits": 211,
      "control_correct": 165,
      "disagreement_selected_correct": 138,
      "disagreement_selection_accuracy": 0.9387755102040817,
      "disagreements": 147,
      "idr1_accuracy": 0.3993558776167472,
      "idr1_correct": 248,
      "oracle_accuracy": 0.45088566827697263,
      "oracle_correct": 280,
      "order_consistency": 1.0,
      "order_consistent": 621,
      "selected_accuracy": 0.4363929146537842,
      "selected_correct": 271,
      "total": 621
    },
    "mbpp": {
      "control_accuracy": 0.7575757575757576,
      "control_commit_rate": 0.06060606060606061,
      "control_commits": 2,
      "control_correct": 25,
      "disagreement_selected_correct": 1,
      "disagreement_selection_accuracy": 1.0,
      "disagreements": 1,
      "idr1_accuracy": 0.7878787878787878,
      "idr1_correct": 26,
      "oracle_accuracy": 0.7878787878787878,
      "oracle_correct": 26,
      "order_consistency": 1.0,
      "order_consistent": 33,
      "selected_accuracy": 0.7878787878787878,
      "selected_correct": 26,
      "total": 33
    },
    "overall": {
      "control_accuracy": 0.38702111024237684,
      "control_commit_rate": 0.47849882720875686,
      "control_commits": 612,
      "control_correct": 495,
      "disagreement_selected_correct": 202,
      "disagreement_selection_accuracy": 0.9099099099099099,
      "disagreements": 222,
      "idr1_accuracy": 0.48866301798279904,
      "idr1_correct": 625,
      "oracle_accuracy": 0.524628616106333,
      "oracle_correct": 671,
      "order_consistency": 1.0,
      "order_consistent": 1279,
      "selected_accuracy": 0.5089913995308835,
      "selected_correct": 651,
      "total": 1279
    }
  },
  "holdout_gate": {
    "code_at_least_24": true,
    "exact_order_consistency": true,
    "logic_at_least_349": true,
    "math_at_least_255": true,
    "net_positive_over_idr1": true,
    "overall_at_least_646": true
  },
  "holdout_gate_pass": true,
  "holdout_maximum_swap_error": 0.0,
  "holdout_prompt_truncated": 1,
  "inference_fields": [
    "question",
    "candidate_a",
    "candidate_b"
  ],
  "max_sequence_length": 3072,
  "model_loader": "multimodal",
  "model_revision": "c202236235762e1c871ad0ccb60c8ee5ba337b9a",
  "model_root": "/lustre/fs1/home/sa305415/shohin/artifacts/external/qwen3.5-9b-c202236",
  "pair_presentations": 1024,
  "pairs": "/lustre/fs1/home/sa305415/shohin/artifacts/product_reasoning/aqc1_pairs_192a7f7_r1/pairs.jsonl",
  "pairs_sha256": "c78bbf075e8b9f04b1e5b460c60bbf4771167065a48d492b124052b94228266d",
  "peak_gpu_memory_bytes": 33375485440,
  "projection_width": 256,
  "protected_adapter_sha256_after": "854a7cc44fbc2b54418f4e5bd09b7efeed0da44fc9ce217b0bb6b1997b722971",
  "protected_adapter_unchanged": true,
  "schema": "shohin-aqc1-commit-report-v1",
  "seed": 2026080822,
  "status": "complete",
  "strata_counts": {
    "bbh_logic:both_correct": 246,
    "bbh_logic:both_wrong": 209,
    "bbh_logic:control_only": 14,
    "bbh_logic:idr1_only": 51,
    "math500:both_correct": 94,
    "math500:both_wrong": 296,
    "math500:control_only": 23,
    "math500:idr1_only": 84,
    "mbpp:both_correct": 11,
    "mbpp:both_wrong": 11,
    "mbpp:idr1_only": 2
  },
  "task_or_benchmark_label_at_inference": false,
  "trace": [
    {
      "gradient_norm": 0.39291074872016907,
      "pairs_per_second": 1.3570643407557397,
      "presentations": 8,
      "prompt_truncated": 0,
      "update": 1
    },
    {
      "gradient_norm": 0.8581412434577942,
      "pairs_per_second": 1.7794829262074512,
      "presentations": 64,
      "prompt_truncated": 0,
      "update": 8
    },
    {
      "gradient_norm": 1.8564153909683228,
      "pairs_per_second": 1.791825555722096,
      "presentations": 128,
      "prompt_truncated": 0,
      "update": 16
    },
    {
      "gradient_norm": 0.3722193241119385,
      "pairs_per_second": 1.7944444930421526,
      "presentations": 192,
      "prompt_truncated": 0,
      "update": 24
    },
    {
      "gradient_norm": 0.6931060552597046,
      "pairs_per_second": 1.7979998170545917,
      "presentations": 256,
      "prompt_truncated": 0,
      "update": 32
    },
    {
      "gradient_norm": 0.6766666769981384,
      "pairs_per_second": 1.7925782089015454,
      "presentations": 320,
      "prompt_truncated": 0,
      "update": 40
    },
    {
      "gradient_norm": 1.0623148679733276,
      "pairs_per_second": 1.8030330572561708,
      "presentations": 384,
      "prompt_truncated": 0,
      "update": 48
    },
    {
      "gradient_norm": 0.7789835929870605,
      "pairs_per_second": 1.8001597643309366,
      "presentations": 448,
      "prompt_truncated": 0,
      "update": 56
    },
    {
      "gradient_norm": 1.8092728853225708,
      "pairs_per_second": 1.8060116852432366,
      "presentations": 512,
      "prompt_truncated": 0,
      "update": 64
    },
    {
      "gradient_norm": 2.3108482360839844,
      "pairs_per_second": 1.8072443301320205,
      "presentations": 576,
      "prompt_truncated": 0,
      "update": 72
    },
    {
      "gradient_norm": 1.185867190361023,
      "pairs_per_second": 1.80962018827459,
      "presentations": 640,
      "prompt_truncated": 0,
      "update": 80
    },
    {
      "gradient_norm": 2.922379493713379,
      "pairs_per_second": 1.8138705167648943,
      "presentations": 704,
      "prompt_truncated": 0,
      "update": 88
    },
    {
      "gradient_norm": 2.359585762023926,
      "pairs_per_second": 1.8128288321734765,
      "presentations": 768,
      "prompt_truncated": 0,
      "update": 96
    },
    {
      "gradient_norm": 1.506990671157837,
      "pairs_per_second": 1.8113371695810507,
      "presentations": 832,
      "prompt_truncated": 0,
      "update": 104
    },
    {
      "gradient_norm": 1.0032193660736084,
      "pairs_per_second": 1.8127459156063777,
      "presentations": 896,
      "prompt_truncated": 0,
      "update": 112
    },
    {
      "gradient_norm": 0.6612640023231506,
      "pairs_per_second": 1.8148904737337392,
      "presentations": 960,
      "prompt_truncated": 0,
      "update": 120
    },
    {
      "gradient_norm": 1.3720777034759521,
      "pairs_per_second": 1.8152467493256477,
      "presentations": 1024,
      "prompt_truncated": 0,
      "update": 128
    }
  ],
  "training_prompt_truncated": 0,
  "updates": 128
}
