{
  "run": "smoke_vm_echo_masked",
  "args": {
    "run_name": "smoke_vm_echo_masked",
    "model_name": "Qwen/Qwen3-4B",
    "seed": 32,
    "device": "",
    "seed_train_size": 16,
    "unlabeled_train_size": 16,
    "full_supervised_size": 16,
    "val_size": 16,
    "fresh_size": 16,
    "hard_size": 16,
    "max_arith_steps": 3,
    "qwen_batch_size": 8,
    "d_model": 128,
    "layers": 1,
    "heads": 4,
    "dropout": 0.1,
    "batch_size": 8,
    "lr": 0.0005,
    "expert_lr": 0.00025,
    "answer_weight": 0.2,
    "arms": [
      "baseline",
      "vm_echo"
    ],
    "echo_weight": 0.35,
    "echo_trace_weight": 1.0,
    "echo_depth_weight": 0.4,
    "echo_final_weight": 0.5,
    "echo_valid_weight": 0.2,
    "seed_epochs": 1,
    "expert_rounds": 1,
    "expert_epochs": 1,
    "full_supervised_epochs": 1,
    "search_topk": 2,
    "max_two_arg_pairs": 2
  },
  "results": [
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_seed_supervised",
      "split": "val_mixed",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_seed_supervised",
      "split": "fresh_standard",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.010869565217391304,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_seed_supervised",
      "split": "fresh_paraphrase",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_seed_supervised",
      "split": "fresh_paired",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.07317073170731707,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_seed_supervised",
      "split": "hard_composition",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.008620689655172414,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_expert_round_1",
      "split": "val_mixed",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_expert_round_1",
      "split": "fresh_standard",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.010869565217391304,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_expert_round_1",
      "split": "fresh_paraphrase",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_expert_round_1",
      "split": "fresh_paired",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.07317073170731707,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_expert_round_1",
      "split": "hard_composition",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.008620689655172414,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_full_supervised",
      "split": "val_mixed",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_full_supervised",
      "split": "fresh_standard",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.010869565217391304,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_full_supervised",
      "split": "fresh_paraphrase",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_full_supervised",
      "split": "fresh_paired",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.07317073170731707,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "baseline",
      "phase": "baseline_full_supervised",
      "split": "hard_composition",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 0.0,
      "echo_trace_top_accuracy": 0.008620689655172414,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_seed_supervised",
      "split": "val_mixed",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_seed_supervised",
      "split": "fresh_standard",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.010869565217391304,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_seed_supervised",
      "split": "fresh_paraphrase",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_seed_supervised",
      "split": "fresh_paired",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.07317073170731707,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_seed_supervised",
      "split": "hard_composition",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.008620689655172414,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_expert_round_1",
      "split": "val_mixed",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0625,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_expert_round_1",
      "split": "fresh_standard",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.125,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.010869565217391304,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_expert_round_1",
      "split": "fresh_paraphrase",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_expert_round_1",
      "split": "fresh_paired",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.07317073170731707,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_expert_round_1",
      "split": "hard_composition",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.008620689655172414,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_full_supervised",
      "split": "val_mixed",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_full_supervised",
      "split": "fresh_standard",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.010869565217391304,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_full_supervised",
      "split": "fresh_paraphrase",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.0,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_full_supervised",
      "split": "fresh_paired",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.07317073170731707,
      "echo_trace_depth_accuracy": 0.0
    },
    {
      "run": "smoke_vm_echo_masked",
      "arm": "vm_echo",
      "phase": "vm_echo_full_supervised",
      "split": "hard_composition",
      "n": 16,
      "direct_accuracy": 0.0,
      "search_accuracy": 0.0,
      "oracle_accuracy": 0.0,
      "program_exact": 0.0,
      "valid_rate": 0.0,
      "direct_valid_rate": 0.0,
      "mean_candidates": 52.0,
      "found_rate": 0.0,
      "gap_recovered": 0.0,
      "echo_final_accuracy": 0.0,
      "echo_valid_accuracy": 1.0,
      "echo_trace_top_accuracy": 0.008620689655172414,
      "echo_trace_depth_accuracy": 0.0
    }
  ]
}