{
  "checkpoints": [
    {
      "adapter_path": "large_artifacts/qwen_onpolicy_repair_compiler/checkpoints/smoke_onpolicy_repair/round01/adapter",
      "heads_path": "large_artifacts/qwen_onpolicy_repair_compiler/checkpoints/smoke_onpolicy_repair/round01/heads.pt",
      "round": 1,
      "run": "smoke_onpolicy_repair"
    }
  ],
  "metadata": {
    "args": {
      "arg_gold_loss_weight": 4.0,
      "arg_repair_loss_weight": 4.0,
      "compiler_checkpoint": "large_artifacts/qwen_onpolicy_repair_compiler/checkpoints/fixed_compiler_step00800",
      "epochs_per_round": 1,
      "eval_batch_size": 2,
      "eval_examples": 4,
      "eval_length": 24,
      "eval_pairs": 4,
      "eval_seed": 72001,
      "executor_loss_weight": 0.0,
      "gold_trace_loss_weight": 0.05,
      "grad_clip": 1.0,
      "gradient_checkpointing": true,
      "init_gold_loss_weight": 4.0,
      "init_repair_loss_weight": 4.0,
      "lr": 2e-05,
      "max_length": 384,
      "max_steps": 24,
      "modulus": 97,
      "onpolicy_rounds": 1,
      "op_gold_loss_weight": 1.0,
      "op_repair_loss_weight": 1.0,
      "qwen_batch_size": 2,
      "repair_eval": true,
      "repair_include_args": true,
      "repair_include_init": true,
      "repair_include_ops": true,
      "repair_max_edits": 1,
      "repair_max_pair_arg_slots": 4,
      "repair_same_step_op_arg": true,
      "repair_topk": 2,
      "repair_trace_loss_weight": 1.0,
      "repair_verifier_mode": "state",
      "run_name": "smoke_onpolicy_repair",
      "seed": 72,
      "state_loss_weight": 0.0,
      "target_mode": "repair_or_gold",
      "train_batch_size": 1,
      "train_examples": 8,
      "train_length": 24,
      "train_template_modes": "standard,paraphrase",
      "val_examples": 4,
      "weight_decay": 0.0
    },
    "compiler_args": {
      "arg_distance_temperature": 2.0,
      "arg_reader_mode": "after_op",
      "arg_window": 8,
      "head_width": 768,
      "max_steps": 24,
      "model_id": "Qwen/Qwen3-4B",
      "modulus": 97,
      "rank_temperature": 1.0
    },
    "cuda_available": true,
    "gpu_name": "NVIDIA RTX 6000 Ada Generation",
    "platform": "Linux-6.8.0-62-generic-x86_64-with-glibc2.39",
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "train_seconds": 15.101984977722168
  },
  "metrics": {
    "fresh_paired_len24": {
      "arg_accuracy": 0.9479166666666666,
      "arg_pos_accuracy": 0.9479166666666666,
      "compiler_pair_program_consistency": 0.5,
      "compiler_pair_state_consistency": 0.5,
      "direct_accuracy": NaN,
      "direct_target_mass": NaN,
      "executor_accuracy": 0.0,
      "executor_pair_answer_consistency": 0.5,
      "executor_pair_both_correct": 0.0,
      "executor_target_mass": 0.010309295263141394,
      "init_accuracy": 1.0,
      "init_pos_accuracy": 1.0,
      "n": 8.0,
      "op_accuracy": 1.0,
      "op_pos_accuracy": 1.0,
      "pair_true_answer_consistency": 1.0,
      "pair_true_length_consistency": 1.0,
      "program_exact": 0.0,
      "repair_avg_candidates": 50.0,
      "repair_avg_verified_candidates": 0.25,
      "repair_changed_fraction": 0.25,
      "repair_executor_accuracy": 0.25,
      "repair_found_fraction": 0.25,
      "repair_pair_answer_consistency": 0.5,
      "repair_pair_both_correct": 0.25,
      "repair_pair_program_consistency": 0.5,
      "repair_pair_state_consistency": 0.5,
      "repair_program_exact": 0.25,
      "repair_state_accuracy": 0.5833333333333334,
      "repair_state_all_exact": 0.25,
      "repair_state_mean_correct_prefix": 14.0,
      "repair_state_prefix_fraction": 0.5833333333333334,
      "state_accuracy": 0.5729166666666666,
      "state_all_exact": 0.0,
      "state_mean_correct_prefix": 13.75,
      "state_prefix_fraction": 0.5729166666666666,
      "variant": "onpolicy_repair_compiler"
    },
    "fresh_paraphrase_len24": {
      "arg_accuracy": 0.9375,
      "arg_pos_accuracy": 0.9375,
      "direct_accuracy": NaN,
      "direct_target_mass": NaN,
      "executor_accuracy": 0.5,
      "executor_target_mass": 0.010309222154319286,
      "init_accuracy": 1.0,
      "init_pos_accuracy": 1.0,
      "n": 4.0,
      "op_accuracy": 1.0,
      "op_pos_accuracy": 1.0,
      "program_exact": 0.5,
      "repair_avg_candidates": 50.0,
      "repair_avg_verified_candidates": 0.5,
      "repair_changed_fraction": 0.0,
      "repair_executor_accuracy": 0.5,
      "repair_found_fraction": 0.5,
      "repair_program_exact": 0.5,
      "repair_state_accuracy": 0.75,
      "repair_state_all_exact": 0.5,
      "repair_state_mean_correct_prefix": 18.0,
      "repair_state_prefix_fraction": 0.75,
      "state_accuracy": 0.75,
      "state_all_exact": 0.5,
      "state_mean_correct_prefix": 18.0,
      "state_prefix_fraction": 0.75,
      "variant": "onpolicy_repair_compiler"
    },
    "fresh_standard_len24": {
      "arg_accuracy": 0.96875,
      "arg_pos_accuracy": 0.9791666666666666,
      "direct_accuracy": NaN,
      "direct_target_mass": NaN,
      "executor_accuracy": 0.5,
      "executor_target_mass": 0.010309374891221523,
      "init_accuracy": 1.0,
      "init_pos_accuracy": 1.0,
      "n": 4.0,
      "op_accuracy": 1.0,
      "op_pos_accuracy": 1.0,
      "program_exact": 0.5,
      "repair_avg_candidates": 50.0,
      "repair_avg_verified_candidates": 0.75,
      "repair_changed_fraction": 0.25,
      "repair_executor_accuracy": 0.75,
      "repair_found_fraction": 0.75,
      "repair_program_exact": 0.75,
      "repair_state_accuracy": 0.8333333333333334,
      "repair_state_all_exact": 0.75,
      "repair_state_mean_correct_prefix": 20.0,
      "repair_state_prefix_fraction": 0.8333333333333334,
      "state_accuracy": 0.59375,
      "state_all_exact": 0.5,
      "state_mean_correct_prefix": 14.25,
      "state_prefix_fraction": 0.59375,
      "variant": "onpolicy_repair_compiler"
    },
    "val_len24": {
      "arg_accuracy": 0.9479166666666666,
      "arg_pos_accuracy": 0.9479166666666666,
      "direct_accuracy": NaN,
      "direct_target_mass": NaN,
      "executor_accuracy": 0.25,
      "executor_target_mass": 0.010309238452464342,
      "init_accuracy": 1.0,
      "init_pos_accuracy": 1.0,
      "n": 4.0,
      "op_accuracy": 1.0,
      "op_pos_accuracy": 1.0,
      "program_exact": 0.25,
      "repair_avg_candidates": 50.0,
      "repair_avg_verified_candidates": 0.25,
      "repair_changed_fraction": 0.0,
      "repair_executor_accuracy": 0.25,
      "repair_found_fraction": 0.25,
      "repair_program_exact": 0.25,
      "repair_state_accuracy": 0.4166666666666667,
      "repair_state_all_exact": 0.25,
      "repair_state_mean_correct_prefix": 10.0,
      "repair_state_prefix_fraction": 0.4166666666666667,
      "state_accuracy": 0.4166666666666667,
      "state_all_exact": 0.25,
      "state_mean_correct_prefix": 10.0,
      "state_prefix_fraction": 0.4166666666666667,
      "variant": "onpolicy_repair_compiler"
    }
  },
  "train_log": [
    {
      "epoch": 0,
      "fresh_paired_len24_arg_accuracy": 0.9479166666666666,
      "fresh_paired_len24_arg_pos_accuracy": 0.9479166666666666,
      "fresh_paired_len24_compiler_pair_program_consistency": 0.75,
      "fresh_paired_len24_compiler_pair_state_consistency": 0.75,
      "fresh_paired_len24_direct_accuracy": NaN,
      "fresh_paired_len24_direct_target_mass": NaN,
      "fresh_paired_len24_executor_accuracy": 0.0,
      "fresh_paired_len24_executor_pair_answer_consistency": 0.75,
      "fresh_paired_len24_executor_pair_both_correct": 0.0,
      "fresh_paired_len24_executor_target_mass": 0.0103092638310045,
      "fresh_paired_len24_init_accuracy": 1.0,
      "fresh_paired_len24_init_pos_accuracy": 1.0,
      "fresh_paired_len24_op_accuracy": 1.0,
      "fresh_paired_len24_op_pos_accuracy": 1.0,
      "fresh_paired_len24_pair_true_answer_consistency": 1.0,
      "fresh_paired_len24_pair_true_length_consistency": 1.0,
      "fresh_paired_len24_program_exact": 0.0,
      "fresh_paired_len24_repair_avg_candidates": 50.0,
      "fresh_paired_len24_repair_avg_verified_candidates": 0.25,
      "fresh_paired_len24_repair_changed_fraction": 0.25,
      "fresh_paired_len24_repair_executor_accuracy": 0.25,
      "fresh_paired_len24_repair_found_fraction": 0.25,
      "fresh_paired_len24_repair_pair_answer_consistency": 0.75,
      "fresh_paired_len24_repair_pair_both_correct": 0.25,
      "fresh_paired_len24_repair_pair_program_consistency": 0.75,
      "fresh_paired_len24_repair_pair_state_consistency": 0.75,
      "fresh_paired_len24_repair_program_exact": 0.25,
      "fresh_paired_len24_repair_state_accuracy": 0.5833333333333334,
      "fresh_paired_len24_repair_state_all_exact": 0.25,
      "fresh_paired_len24_repair_state_mean_correct_prefix": 14.0,
      "fresh_paired_len24_repair_state_prefix_fraction": 0.5833333333333334,
      "fresh_paired_len24_state_accuracy": 0.5729166666666666,
      "fresh_paired_len24_state_all_exact": 0.0,
      "fresh_paired_len24_state_mean_correct_prefix": 13.75,
      "fresh_paired_len24_state_prefix_fraction": 0.5729166666666666,
      "fresh_paraphrase_len24_arg_accuracy": 0.9479166666666666,
      "fresh_paraphrase_len24_arg_pos_accuracy": 0.9479166666666666,
      "fresh_paraphrase_len24_direct_accuracy": NaN,
      "fresh_paraphrase_len24_direct_target_mass": NaN,
      "fresh_paraphrase_len24_executor_accuracy": 0.5,
      "fresh_paraphrase_len24_executor_target_mass": 0.010309302248060703,
      "fresh_paraphrase_len24_init_accuracy": 1.0,
      "fresh_paraphrase_len24_init_pos_accuracy": 1.0,
      "fresh_paraphrase_len24_op_accuracy": 1.0,
      "fresh_paraphrase_len24_op_pos_accuracy": 1.0,
      "fresh_paraphrase_len24_program_exact": 0.5,
      "fresh_paraphrase_len24_repair_avg_candidates": 50.0,
      "fresh_paraphrase_len24_repair_avg_verified_candidates": 0.5,
      "fresh_paraphrase_len24_repair_changed_fraction": 0.0,
      "fresh_paraphrase_len24_repair_executor_accuracy": 0.5,
      "fresh_paraphrase_len24_repair_found_fraction": 0.5,
      "fresh_paraphrase_len24_repair_program_exact": 0.5,
      "fresh_paraphrase_len24_repair_state_accuracy": 0.75,
      "fresh_paraphrase_len24_repair_state_all_exact": 0.5,
      "fresh_paraphrase_len24_repair_state_mean_correct_prefix": 18.0,
      "fresh_paraphrase_len24_repair_state_prefix_fraction": 0.75,
      "fresh_paraphrase_len24_state_accuracy": 0.75,
      "fresh_paraphrase_len24_state_all_exact": 0.5,
      "fresh_paraphrase_len24_state_mean_correct_prefix": 18.0,
      "fresh_paraphrase_len24_state_prefix_fraction": 0.75,
      "fresh_standard_len24_arg_accuracy": 0.96875,
      "fresh_standard_len24_arg_pos_accuracy": 0.9791666666666666,
      "fresh_standard_len24_direct_accuracy": NaN,
      "fresh_standard_len24_direct_target_mass": NaN,
      "fresh_standard_len24_executor_accuracy": 0.5,
      "fresh_standard_len24_executor_target_mass": 0.010309378616511822,
      "fresh_standard_len24_init_accuracy": 1.0,
      "fresh_standard_len24_init_pos_accuracy": 1.0,
      "fresh_standard_len24_op_accuracy": 1.0,
      "fresh_standard_len24_op_pos_accuracy": 1.0,
      "fresh_standard_len24_program_exact": 0.5,
      "fresh_standard_len24_repair_avg_candidates": 50.0,
      "fresh_standard_len24_repair_avg_verified_candidates": 0.75,
      "fresh_standard_len24_repair_changed_fraction": 0.25,
      "fresh_standard_len24_repair_executor_accuracy": 0.75,
      "fresh_standard_len24_repair_found_fraction": 0.75,
      "fresh_standard_len24_repair_program_exact": 0.75,
      "fresh_standard_len24_repair_state_accuracy": 0.8333333333333334,
      "fresh_standard_len24_repair_state_all_exact": 0.75,
      "fresh_standard_len24_repair_state_mean_correct_prefix": 20.0,
      "fresh_standard_len24_repair_state_prefix_fraction": 0.8333333333333334,
      "fresh_standard_len24_state_accuracy": 0.59375,
      "fresh_standard_len24_state_all_exact": 0.5,
      "fresh_standard_len24_state_mean_correct_prefix": 14.25,
      "fresh_standard_len24_state_prefix_fraction": 0.59375,
      "phase": "baseline",
      "round": 0,
      "val_len24_arg_accuracy": 0.9479166666666666,
      "val_len24_arg_pos_accuracy": 0.9479166666666666,
      "val_len24_direct_accuracy": NaN,
      "val_len24_direct_target_mass": NaN,
      "val_len24_executor_accuracy": 0.25,
      "val_len24_executor_target_mass": 0.010309246368706226,
      "val_len24_init_accuracy": 1.0,
      "val_len24_init_pos_accuracy": 1.0,
      "val_len24_op_accuracy": 1.0,
      "val_len24_op_pos_accuracy": 1.0,
      "val_len24_program_exact": 0.25,
      "val_len24_repair_avg_candidates": 50.0,
      "val_len24_repair_avg_verified_candidates": 0.25,
      "val_len24_repair_changed_fraction": 0.0,
      "val_len24_repair_executor_accuracy": 0.25,
      "val_len24_repair_found_fraction": 0.25,
      "val_len24_repair_program_exact": 0.25,
      "val_len24_repair_state_accuracy": 0.4166666666666667,
      "val_len24_repair_state_all_exact": 0.25,
      "val_len24_repair_state_mean_correct_prefix": 10.0,
      "val_len24_repair_state_prefix_fraction": 0.4166666666666667,
      "val_len24_state_accuracy": 0.4166666666666667,
      "val_len24_state_all_exact": 0.25,
      "val_len24_state_mean_correct_prefix": 10.0,
      "val_len24_state_prefix_fraction": 0.4166666666666667
    },
    {
      "epoch": 1,
      "fresh_paired_len24_arg_accuracy": 0.9479166666666666,
      "fresh_paired_len24_arg_pos_accuracy": 0.9479166666666666,
      "fresh_paired_len24_compiler_pair_program_consistency": 0.5,
      "fresh_paired_len24_compiler_pair_state_consistency": 0.5,
      "fresh_paired_len24_direct_accuracy": NaN,
      "fresh_paired_len24_direct_target_mass": NaN,
      "fresh_paired_len24_executor_accuracy": 0.0,
      "fresh_paired_len24_executor_pair_answer_consistency": 0.5,
      "fresh_paired_len24_executor_pair_both_correct": 0.0,
      "fresh_paired_len24_executor_target_mass": 0.010309295263141394,
      "fresh_paired_len24_init_accuracy": 1.0,
      "fresh_paired_len24_init_pos_accuracy": 1.0,
      "fresh_paired_len24_op_accuracy": 1.0,
      "fresh_paired_len24_op_pos_accuracy": 1.0,
      "fresh_paired_len24_pair_true_answer_consistency": 1.0,
      "fresh_paired_len24_pair_true_length_consistency": 1.0,
      "fresh_paired_len24_program_exact": 0.0,
      "fresh_paired_len24_repair_avg_candidates": 50.0,
      "fresh_paired_len24_repair_avg_verified_candidates": 0.25,
      "fresh_paired_len24_repair_changed_fraction": 0.25,
      "fresh_paired_len24_repair_executor_accuracy": 0.25,
      "fresh_paired_len24_repair_found_fraction": 0.25,
      "fresh_paired_len24_repair_pair_answer_consistency": 0.5,
      "fresh_paired_len24_repair_pair_both_correct": 0.25,
      "fresh_paired_len24_repair_pair_program_consistency": 0.5,
      "fresh_paired_len24_repair_pair_state_consistency": 0.5,
      "fresh_paired_len24_repair_program_exact": 0.25,
      "fresh_paired_len24_repair_state_accuracy": 0.5833333333333334,
      "fresh_paired_len24_repair_state_all_exact": 0.25,
      "fresh_paired_len24_repair_state_mean_correct_prefix": 14.0,
      "fresh_paired_len24_repair_state_prefix_fraction": 0.5833333333333334,
      "fresh_paired_len24_state_accuracy": 0.5729166666666666,
      "fresh_paired_len24_state_all_exact": 0.0,
      "fresh_paired_len24_state_mean_correct_prefix": 13.75,
      "fresh_paired_len24_state_prefix_fraction": 0.5729166666666666,
      "fresh_paraphrase_len24_arg_accuracy": 0.9375,
      "fresh_paraphrase_len24_arg_pos_accuracy": 0.9375,
      "fresh_paraphrase_len24_direct_accuracy": NaN,
      "fresh_paraphrase_len24_direct_target_mass": NaN,
      "fresh_paraphrase_len24_executor_accuracy": 0.5,
      "fresh_paraphrase_len24_executor_target_mass": 0.010309222154319286,
      "fresh_paraphrase_len24_init_accuracy": 1.0,
      "fresh_paraphrase_len24_init_pos_accuracy": 1.0,
      "fresh_paraphrase_len24_op_accuracy": 1.0,
      "fresh_paraphrase_len24_op_pos_accuracy": 1.0,
      "fresh_paraphrase_len24_program_exact": 0.5,
      "fresh_paraphrase_len24_repair_avg_candidates": 50.0,
      "fresh_paraphrase_len24_repair_avg_verified_candidates": 0.5,
      "fresh_paraphrase_len24_repair_changed_fraction": 0.0,
      "fresh_paraphrase_len24_repair_executor_accuracy": 0.5,
      "fresh_paraphrase_len24_repair_found_fraction": 0.5,
      "fresh_paraphrase_len24_repair_program_exact": 0.5,
      "fresh_paraphrase_len24_repair_state_accuracy": 0.75,
      "fresh_paraphrase_len24_repair_state_all_exact": 0.5,
      "fresh_paraphrase_len24_repair_state_mean_correct_prefix": 18.0,
      "fresh_paraphrase_len24_repair_state_prefix_fraction": 0.75,
      "fresh_paraphrase_len24_state_accuracy": 0.75,
      "fresh_paraphrase_len24_state_all_exact": 0.5,
      "fresh_paraphrase_len24_state_mean_correct_prefix": 18.0,
      "fresh_paraphrase_len24_state_prefix_fraction": 0.75,
      "fresh_standard_len24_arg_accuracy": 0.96875,
      "fresh_standard_len24_arg_pos_accuracy": 0.9791666666666666,
      "fresh_standard_len24_direct_accuracy": NaN,
      "fresh_standard_len24_direct_target_mass": NaN,
      "fresh_standard_len24_executor_accuracy": 0.5,
      "fresh_standard_len24_executor_target_mass": 0.010309374891221523,
      "fresh_standard_len24_init_accuracy": 1.0,
      "fresh_standard_len24_init_pos_accuracy": 1.0,
      "fresh_standard_len24_op_accuracy": 1.0,
      "fresh_standard_len24_op_pos_accuracy": 1.0,
      "fresh_standard_len24_program_exact": 0.5,
      "fresh_standard_len24_repair_avg_candidates": 50.0,
      "fresh_standard_len24_repair_avg_verified_candidates": 0.75,
      "fresh_standard_len24_repair_changed_fraction": 0.25,
      "fresh_standard_len24_repair_executor_accuracy": 0.75,
      "fresh_standard_len24_repair_found_fraction": 0.75,
      "fresh_standard_len24_repair_program_exact": 0.75,
      "fresh_standard_len24_repair_state_accuracy": 0.8333333333333334,
      "fresh_standard_len24_repair_state_all_exact": 0.75,
      "fresh_standard_len24_repair_state_mean_correct_prefix": 20.0,
      "fresh_standard_len24_repair_state_prefix_fraction": 0.8333333333333334,
      "fresh_standard_len24_state_accuracy": 0.59375,
      "fresh_standard_len24_state_all_exact": 0.5,
      "fresh_standard_len24_state_mean_correct_prefix": 14.25,
      "fresh_standard_len24_state_prefix_fraction": 0.59375,
      "phase": "train",
      "round": 1,
      "target_active_fraction": 1.0,
      "target_avg_candidates": 50.0,
      "target_avg_verified": 1.125,
      "target_changed_fraction": 0.0,
      "target_repair_fraction": 0.625,
      "target_source_base_fallback": 0,
      "target_source_gold_fallback": 3,
      "target_source_repair": 5,
      "target_source_skipped": 0,
      "train_executor_loss": 0.0,
      "train_gold_trace_loss": 2.6203933656215668,
      "train_loss": 2.75141304731369,
      "train_repair_arg_loss": 0.566822960972786,
      "train_repair_init_loss": 0.0007101649534888566,
      "train_repair_op_loss": 0.3502608798444271,
      "train_state_loss": 0.0,
      "val_len24_arg_accuracy": 0.9479166666666666,
      "val_len24_arg_pos_accuracy": 0.9479166666666666,
      "val_len24_direct_accuracy": NaN,
      "val_len24_direct_target_mass": NaN,
      "val_len24_executor_accuracy": 0.25,
      "val_len24_executor_target_mass": 0.010309238452464342,
      "val_len24_init_accuracy": 1.0,
      "val_len24_init_pos_accuracy": 1.0,
      "val_len24_op_accuracy": 1.0,
      "val_len24_op_pos_accuracy": 1.0,
      "val_len24_program_exact": 0.25,
      "val_len24_repair_avg_candidates": 50.0,
      "val_len24_repair_avg_verified_candidates": 0.25,
      "val_len24_repair_changed_fraction": 0.0,
      "val_len24_repair_executor_accuracy": 0.25,
      "val_len24_repair_found_fraction": 0.25,
      "val_len24_repair_program_exact": 0.25,
      "val_len24_repair_state_accuracy": 0.4166666666666667,
      "val_len24_repair_state_all_exact": 0.25,
      "val_len24_repair_state_mean_correct_prefix": 10.0,
      "val_len24_repair_state_prefix_fraction": 0.4166666666666667,
      "val_len24_state_accuracy": 0.4166666666666667,
      "val_len24_state_all_exact": 0.25,
      "val_len24_state_mean_correct_prefix": 10.0,
      "val_len24_state_prefix_fraction": 0.4166666666666667
    }
  ]
}