{
  "arms": [
    {
      "adapter_dir": null,
      "arm_name": "base_t09_k4",
      "count": 16,
      "experiment": "qwen35_4b_passk_coverage_rl",
      "max_new_tokens": 220,
      "offset": 0,
      "records": {
        "coverage": 0.5,
        "forward_tokens": 14193,
        "hidden_pass_candidates": 23,
        "mean_candidates": 4.0,
        "mean_distinct_functional_rate": 0.46875,
        "mean_distinct_program_rate": 0.953125,
        "pass1_proxy": 0.375,
        "task_count": 16,
        "visible_candidates": 33,
        "visible_coverage": 0.75
      },
      "round_name": "pilot_base_t09",
      "samples_per_task": 4,
      "seed": 20260633,
      "split": "test",
      "temperatures": "0.9",
      "top_p": 0.95
    },
    {
      "adapter_dir": null,
      "arm_name": "base_hot_k4",
      "count": 16,
      "experiment": "qwen35_4b_passk_coverage_rl",
      "max_new_tokens": 220,
      "offset": 0,
      "records": {
        "coverage": 0.6875,
        "forward_tokens": 14959,
        "hidden_pass_candidates": 27,
        "mean_candidates": 4.0,
        "mean_distinct_functional_rate": 0.484375,
        "mean_distinct_program_rate": 0.96875,
        "pass1_proxy": 0.4375,
        "task_count": 16,
        "visible_candidates": 35,
        "visible_coverage": 0.75
      },
      "round_name": "pilot_base_hot",
      "samples_per_task": 4,
      "seed": 20260633,
      "split": "test",
      "temperatures": "1.0,1.2",
      "top_p": 0.95
    },
    {
      "adapter_dir": "/workspace/large_artifacts/qwen35_4b_passk_coverage_rl/pilot_unsat_passk_rl_adapter",
      "arm_name": "passk_rl_unsat_k4",
      "count": 16,
      "experiment": "qwen35_4b_passk_coverage_rl",
      "max_new_tokens": 220,
      "offset": 0,
      "records": {
        "coverage": 0.4375,
        "forward_tokens": 13650,
        "hidden_pass_candidates": 24,
        "mean_candidates": 4.0,
        "mean_distinct_functional_rate": 0.40625,
        "mean_distinct_program_rate": 0.96875,
        "pass1_proxy": 0.3125,
        "task_count": 16,
        "visible_candidates": 39,
        "visible_coverage": 0.6875
      },
      "round_name": "pilot_passk_rl_eval",
      "samples_per_task": 4,
      "seed": 20260633,
      "split": "test",
      "temperatures": "0.9",
      "top_p": 0.95
    },
    {
      "adapter_dir": null,
      "arm_name": "base_t09_k8_sample_more",
      "count": 16,
      "experiment": "qwen35_4b_passk_coverage_rl",
      "max_new_tokens": 220,
      "offset": 0,
      "records": {
        "coverage": 0.625,
        "forward_tokens": 28487,
        "hidden_pass_candidates": 49,
        "mean_candidates": 8.0,
        "mean_distinct_functional_rate": 0.2734375,
        "mean_distinct_program_rate": 0.921875,
        "pass1_proxy": 0.375,
        "task_count": 16,
        "visible_candidates": 66,
        "visible_coverage": 0.875
      },
      "round_name": "pilot_base_t09_k8",
      "samples_per_task": 8,
      "seed": 20260633,
      "split": "test",
      "temperatures": "0.9",
      "top_p": 0.95
    }
  ],
  "training": {
    "attempts": 16,
    "elapsed_s": 516.71,
    "experiment": "qwen35_4b_passk_coverage_rl",
    "kl_coef": 0.05,
    "learning_rate": 5e-05,
    "max_attempts": 40,
    "max_steps": 6,
    "method": "online_passk_rl",
    "metrics": [
      {
        "adv_abs_mean": 0.0,
        "adv_mean": 0.0,
        "attempt": 1,
        "group_size": 8,
        "kl_proxy": 0.0,
        "loss": 0.0,
        "mean_completion_tokens": 180.0,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 0.0,
        "pg_loss": -0.0,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 0,
        "task_id": 601,
        "visible_count": 0
      },
      {
        "adv_abs_mean": 0.17500001192092896,
        "adv_mean": -0.17500001192092896,
        "attempt": 2,
        "group_size": 8,
        "kl_proxy": 0.0,
        "loss": -0.1954595446586609,
        "mean_completion_tokens": 105.5,
        "mean_gen_tokens_for_logprob": 177.625,
        "passk_utility": 0.0,
        "pg_loss": -0.1954595446586609,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 0,
        "task_id": 602,
        "visible_count": 7
      },
      {
        "adv_abs_mean": 0.02500000037252903,
        "adv_mean": -0.02500000037252903,
        "attempt": 3,
        "group_size": 8,
        "kl_proxy": 0.0,
        "loss": -0.013361622579395771,
        "mean_completion_tokens": 166.25,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 0.0,
        "pg_loss": -0.013361622579395771,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 0,
        "task_id": 603,
        "visible_count": 1
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 4,
        "group_size": 8,
        "kl_proxy": 0.0,
        "loss": 0.2013738751411438,
        "mean_completion_tokens": 76.5,
        "mean_gen_tokens_for_logprob": 159.75,
        "passk_utility": 1.0,
        "pg_loss": 0.2013738751411438,
        "positive_count": 7,
        "skip_reason": "",
        "skipped": false,
        "step": 1,
        "task_id": 604,
        "visible_count": 7
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 5,
        "group_size": 8,
        "kl_proxy": 0.04197895526885986,
        "loss": 0.10074320435523987,
        "mean_completion_tokens": 110.5,
        "mean_gen_tokens_for_logprob": 177.0,
        "passk_utility": 1.0,
        "pg_loss": 0.09864425659179688,
        "positive_count": 8,
        "skip_reason": "saturated_positive",
        "skipped": true,
        "step": 1,
        "task_id": 605,
        "visible_count": 8
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 6,
        "group_size": 8,
        "kl_proxy": 0.08403637260198593,
        "loss": 0.1418113261461258,
        "mean_completion_tokens": 40.375,
        "mean_gen_tokens_for_logprob": 113.0,
        "passk_utility": 1.0,
        "pg_loss": 0.13760951161384583,
        "positive_count": 8,
        "skip_reason": "saturated_positive",
        "skipped": true,
        "step": 1,
        "task_id": 606,
        "visible_count": 8
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 7,
        "group_size": 8,
        "kl_proxy": 0.025866957381367683,
        "loss": 0.09162773191928864,
        "mean_completion_tokens": 117.25,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 1.0,
        "pg_loss": 0.09033438563346863,
        "positive_count": 6,
        "skip_reason": "",
        "skipped": false,
        "step": 2,
        "task_id": 607,
        "visible_count": 6
      },
      {
        "adv_abs_mean": 0.02500000037252903,
        "adv_mean": -0.02500000037252903,
        "attempt": 8,
        "group_size": 8,
        "kl_proxy": 3.821446171059506e-06,
        "loss": -0.0077784135937690735,
        "mean_completion_tokens": 177.125,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 0.0,
        "pg_loss": -0.00777860451489687,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 2,
        "task_id": 608,
        "visible_count": 1
      },
      {
        "adv_abs_mean": 0.02500000037252903,
        "adv_mean": -0.02500000037252903,
        "attempt": 9,
        "group_size": 8,
        "kl_proxy": 0.20229381322860718,
        "loss": -0.02697703242301941,
        "mean_completion_tokens": 71.125,
        "mean_gen_tokens_for_logprob": 164.0,
        "passk_utility": 0.0,
        "pg_loss": -0.03709172457456589,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 2,
        "task_id": 609,
        "visible_count": 1
      },
      {
        "adv_abs_mean": 0.1160714328289032,
        "adv_mean": 0.1160714328289032,
        "attempt": 10,
        "group_size": 8,
        "kl_proxy": 0.07758591324090958,
        "loss": 0.06510908156633377,
        "mean_completion_tokens": 86.125,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 0.9285714285714286,
        "pg_loss": 0.06122978776693344,
        "positive_count": 3,
        "skip_reason": "",
        "skipped": false,
        "step": 3,
        "task_id": 610,
        "visible_count": 3
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 11,
        "group_size": 8,
        "kl_proxy": 0.07719863951206207,
        "loss": 0.10137037187814713,
        "mean_completion_tokens": 109.375,
        "mean_gen_tokens_for_logprob": 178.25,
        "passk_utility": 1.0,
        "pg_loss": 0.0975104421377182,
        "positive_count": 7,
        "skip_reason": "",
        "skipped": false,
        "step": 4,
        "task_id": 611,
        "visible_count": 7
      },
      {
        "adv_abs_mean": 0.05000000074505806,
        "adv_mean": -0.05000000074505806,
        "attempt": 12,
        "group_size": 8,
        "kl_proxy": 2.8763370210072026e-05,
        "loss": -0.008809228427708149,
        "mean_completion_tokens": 160.375,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 0.0,
        "pg_loss": -0.008810666389763355,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 4,
        "task_id": 612,
        "visible_count": 2
      },
      {
        "adv_abs_mean": 0.13750000298023224,
        "adv_mean": 0.11250000447034836,
        "attempt": 13,
        "group_size": 8,
        "kl_proxy": 0.01994357630610466,
        "loss": 0.036828234791755676,
        "mean_completion_tokens": 106.625,
        "mean_gen_tokens_for_logprob": 156.0,
        "passk_utility": 1.0,
        "pg_loss": 0.03583105653524399,
        "positive_count": 6,
        "skip_reason": "",
        "skipped": false,
        "step": 5,
        "task_id": 613,
        "visible_count": 7
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 14,
        "group_size": 8,
        "kl_proxy": 0.0849635899066925,
        "loss": 0.07477525621652603,
        "mean_completion_tokens": 84.875,
        "mean_gen_tokens_for_logprob": 144.0,
        "passk_utility": 1.0,
        "pg_loss": 0.0705270767211914,
        "positive_count": 8,
        "skip_reason": "saturated_positive",
        "skipped": true,
        "step": 5,
        "task_id": 614,
        "visible_count": 8
      },
      {
        "adv_abs_mean": 0.0,
        "adv_mean": 0.0,
        "attempt": 15,
        "group_size": 8,
        "kl_proxy": 5.106851858727168e-06,
        "loss": 2.553425986207003e-07,
        "mean_completion_tokens": 175.375,
        "mean_gen_tokens_for_logprob": 181.0,
        "passk_utility": 0.0,
        "pg_loss": -0.0,
        "positive_count": 0,
        "skip_reason": "zero_positive",
        "skipped": true,
        "step": 5,
        "task_id": 615,
        "visible_count": 0
      },
      {
        "adv_abs_mean": 0.125,
        "adv_mean": 0.125,
        "attempt": 16,
        "group_size": 8,
        "kl_proxy": 0.006418587639927864,
        "loss": 0.04374063387513161,
        "mean_completion_tokens": 84.875,
        "mean_gen_tokens_for_logprob": 130.0,
        "passk_utility": 1.0,
        "pg_loss": 0.04341970384120941,
        "positive_count": 7,
        "skip_reason": "",
        "skipped": false,
        "step": 6,
        "task_id": 616,
        "visible_count": 7
      }
    ],
    "negative_weight": 0.2,
    "rollouts_per_task": 8,
    "skip_saturated_positive": true,
    "skip_zero_positive": true,
    "target_k": 4,
    "temperature": 0.9,
    "top_p": 0.95,
    "train_count": 32,
    "train_split": "train",
    "updates": 6
  }
}
