{
  "args": {
    "model_id": "Qwen/Qwen3.5-4B",
    "text_model_fallback": "Qwen/Qwen3-4B",
    "model_class": "auto",
    "load_in_4bit": true,
    "torch_dtype": "bf16",
    "device_map": "auto",
    "modulus": 97,
    "max_steps": 24,
    "train_min_len": 1,
    "train_max_len": 4,
    "eval_lengths": "4,12,16,24",
    "train_size": 1024,
    "eval_size": 128,
    "feature_batch_size": 8,
    "train_batch_size": 128,
    "eval_batch_size": 128,
    "train_steps": 800,
    "eval_every": 200,
    "lr": 0.001,
    "weight_decay": 0.01,
    "grad_clip": 1.0,
    "head_width": 512,
    "trace_loss_weight": 1.0,
    "executor_loss_weight": 1.0,
    "variants": "compiler_trace",
    "max_length": 512,
    "seed": 7,
    "output_dir": "experiments/qwen_structured_bridge/runs/scale_qwen35_length24_trace",
    "checkpoint_dir": "",
    "verbose": false
  },
  "metadata": {
    "python": "3.12.3",
    "platform": "Linux-6.8.0-62-generic-x86_64-with-glibc2.39",
    "torch": "2.8.0+cu128",
    "cuda_available": true,
    "transformers_loader": "AutoModelForMultimodalLM",
    "peft_installed": true,
    "gpu_name": "NVIDIA RTX 6000 Ada Generation",
    "gpu_vram_gb": 47.383
  },
  "hidden_dim": 2560,
  "dataset": {
    "train_size": 1024,
    "eval_size": 128,
    "train_lengths": [
      1,
      4
    ],
    "eval_lengths": [
      4,
      12,
      16,
      24
    ]
  },
  "variants": {
    "compiler_trace": {
      "variant": "compiler_trace",
      "train_seconds": 52.61953115463257,
      "checkpoints": [
        "large_artifacts/qwen_structured_bridge/checkpoints/scale_qwen35_length24_trace/compiler_trace/checkpoint_final.pt"
      ],
      "train_log": [
        {
          "step": 1,
          "loss": 14.824259757995605,
          "executor_loss": 4.579751014709473,
          "init_loss": 4.582459449768066,
          "op_loss": 1.09193754196167,
          "arg_loss": 4.570111274719238,
          "len4_direct_accuracy": NaN,
          "len4_direct_target_mass": NaN,
          "len4_executor_accuracy": 0.0234375,
          "len4_executor_target_mass": 0.010300758294761181,
          "len4_init_accuracy": 0.015625,
          "len4_op_accuracy": 0.3125,
          "len4_arg_accuracy": 0.18359375,
          "len4_program_exact": 0.0,
          "len12_direct_accuracy": NaN,
          "len12_direct_target_mass": NaN,
          "len12_executor_accuracy": 0.0,
          "len12_executor_target_mass": 0.01030709408223629,
          "len12_init_accuracy": 0.0546875,
          "len12_op_accuracy": 0.33984375,
          "len12_arg_accuracy": 0.18229166666666666,
          "len12_program_exact": 0.0,
          "len16_direct_accuracy": NaN,
          "len16_direct_target_mass": NaN,
          "len16_executor_accuracy": 0.015625,
          "len16_executor_target_mass": 0.010309652425348759,
          "len16_init_accuracy": 0.0234375,
          "len16_op_accuracy": 0.32177734375,
          "len16_arg_accuracy": 0.17626953125,
          "len16_program_exact": 0.0,
          "len24_direct_accuracy": NaN,
          "len24_direct_target_mass": NaN,
          "len24_executor_accuracy": 0.015625,
          "len24_executor_target_mass": 0.01030733622610569,
          "len24_init_accuracy": 0.0390625,
          "len24_op_accuracy": 0.3369140625,
          "len24_arg_accuracy": 0.1669921875,
          "len24_program_exact": 0.0
        },
        {
          "step": 200,
          "loss": 0.0024112588725984097,
          "executor_loss": 0.0013416714500635862,
          "init_loss": 0.0008768717525526881,
          "op_loss": 1.3820648746332154e-05,
          "arg_loss": 0.00017889501759782434,
          "len4_direct_accuracy": NaN,
          "len4_direct_target_mass": NaN,
          "len4_executor_accuracy": 1.0,
          "len4_executor_target_mass": 0.9971430897712708,
          "len4_init_accuracy": 1.0,
          "len4_op_accuracy": 1.0,
          "len4_arg_accuracy": 1.0,
          "len4_program_exact": 1.0,
          "len12_direct_accuracy": NaN,
          "len12_direct_target_mass": NaN,
          "len12_executor_accuracy": 0.953125,
          "len12_executor_target_mass": 0.9123052358627319,
          "len12_init_accuracy": 1.0,
          "len12_op_accuracy": 0.99609375,
          "len12_arg_accuracy": 1.0,
          "len12_program_exact": 0.953125,
          "len16_direct_accuracy": NaN,
          "len16_direct_target_mass": NaN,
          "len16_executor_accuracy": 0.9453125,
          "len16_executor_target_mass": 0.9071547985076904,
          "len16_init_accuracy": 1.0,
          "len16_op_accuracy": 0.99609375,
          "len16_arg_accuracy": 1.0,
          "len16_program_exact": 0.9453125,
          "len24_direct_accuracy": NaN,
          "len24_direct_target_mass": NaN,
          "len24_executor_accuracy": 0.875,
          "len24_executor_target_mass": 0.8167543411254883,
          "len24_init_accuracy": 1.0,
          "len24_op_accuracy": 0.9947916666666666,
          "len24_arg_accuracy": 1.0,
          "len24_program_exact": 0.875
        },
        {
          "step": 400,
          "loss": 0.0007649306789971888,
          "executor_loss": 0.00042066589230671525,
          "init_loss": 0.00028347905026748776,
          "op_loss": 4.429532509675482e-06,
          "arg_loss": 5.635617708321661e-05,
          "len4_direct_accuracy": NaN,
          "len4_direct_target_mass": NaN,
          "len4_executor_accuracy": 1.0,
          "len4_executor_target_mass": 0.9983950257301331,
          "len4_init_accuracy": 1.0,
          "len4_op_accuracy": 1.0,
          "len4_arg_accuracy": 1.0,
          "len4_program_exact": 1.0,
          "len12_direct_accuracy": NaN,
          "len12_direct_target_mass": NaN,
          "len12_executor_accuracy": 0.9375,
          "len12_executor_target_mass": 0.9338151216506958,
          "len12_init_accuracy": 1.0,
          "len12_op_accuracy": 0.9947916666666666,
          "len12_arg_accuracy": 1.0,
          "len12_program_exact": 0.9375,
          "len16_direct_accuracy": NaN,
          "len16_direct_target_mass": NaN,
          "len16_executor_accuracy": 0.9296875,
          "len16_executor_target_mass": 0.9062188863754272,
          "len16_init_accuracy": 1.0,
          "len16_op_accuracy": 0.99462890625,
          "len16_arg_accuracy": 1.0,
          "len16_program_exact": 0.9296875,
          "len24_direct_accuracy": NaN,
          "len24_direct_target_mass": NaN,
          "len24_executor_accuracy": 0.8828125,
          "len24_executor_target_mass": 0.8106086254119873,
          "len24_init_accuracy": 1.0,
          "len24_op_accuracy": 0.9944661458333334,
          "len24_arg_accuracy": 1.0,
          "len24_program_exact": 0.8828125
        },
        {
          "step": 600,
          "loss": 0.0003765666042454541,
          "executor_loss": 0.00020843450329266489,
          "init_loss": 0.00014070156612433493,
          "op_loss": 4.383528562357242e-07,
          "arg_loss": 2.6992169296136126e-05,
          "len4_direct_accuracy": NaN,
          "len4_direct_target_mass": NaN,
          "len4_executor_accuracy": 1.0,
          "len4_executor_target_mass": 0.9988540410995483,
          "len4_init_accuracy": 1.0,
          "len4_op_accuracy": 1.0,
          "len4_arg_accuracy": 1.0,
          "len4_program_exact": 1.0,
          "len12_direct_accuracy": NaN,
          "len12_direct_target_mass": NaN,
          "len12_executor_accuracy": 0.9375,
          "len12_executor_target_mass": 0.935754656791687,
          "len12_init_accuracy": 1.0,
          "len12_op_accuracy": 0.9947916666666666,
          "len12_arg_accuracy": 1.0,
          "len12_program_exact": 0.9375,
          "len16_direct_accuracy": NaN,
          "len16_direct_target_mass": NaN,
          "len16_executor_accuracy": 0.9296875,
          "len16_executor_target_mass": 0.9127329587936401,
          "len16_init_accuracy": 1.0,
          "len16_op_accuracy": 0.99560546875,
          "len16_arg_accuracy": 1.0,
          "len16_program_exact": 0.9296875,
          "len24_direct_accuracy": NaN,
          "len24_direct_target_mass": NaN,
          "len24_executor_accuracy": 0.875,
          "len24_executor_target_mass": 0.8218545317649841,
          "len24_init_accuracy": 1.0,
          "len24_op_accuracy": 0.9944661458333334,
          "len24_arg_accuracy": 1.0,
          "len24_program_exact": 0.875
        },
        {
          "step": 800,
          "loss": 0.00023410184076055884,
          "executor_loss": 0.00012916112609673291,
          "init_loss": 8.826862904243171e-05,
          "op_loss": 1.0299484074494103e-06,
          "arg_loss": 1.56421465362655e-05,
          "len4_direct_accuracy": NaN,
          "len4_direct_target_mass": NaN,
          "len4_executor_accuracy": 1.0,
          "len4_executor_target_mass": 0.9990867376327515,
          "len4_init_accuracy": 1.0,
          "len4_op_accuracy": 1.0,
          "len4_arg_accuracy": 1.0,
          "len4_program_exact": 1.0,
          "len12_direct_accuracy": NaN,
          "len12_direct_target_mass": NaN,
          "len12_executor_accuracy": 0.9375,
          "len12_executor_target_mass": 0.9364005327224731,
          "len12_init_accuracy": 1.0,
          "len12_op_accuracy": 0.9947916666666666,
          "len12_arg_accuracy": 1.0,
          "len12_program_exact": 0.9375,
          "len16_direct_accuracy": NaN,
          "len16_direct_target_mass": NaN,
          "len16_executor_accuracy": 0.921875,
          "len16_executor_target_mass": 0.9153636693954468,
          "len16_init_accuracy": 1.0,
          "len16_op_accuracy": 0.99560546875,
          "len16_arg_accuracy": 0.99951171875,
          "len16_program_exact": 0.921875,
          "len24_direct_accuracy": NaN,
          "len24_direct_target_mass": NaN,
          "len24_executor_accuracy": 0.875,
          "len24_executor_target_mass": 0.8271066546440125,
          "len24_init_accuracy": 1.0,
          "len24_op_accuracy": 0.9947916666666666,
          "len24_arg_accuracy": 1.0,
          "len24_program_exact": 0.875
        }
      ],
      "final_metrics": {
        "len4": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 1.0,
          "executor_target_mass": 0.9990867376327515,
          "init_accuracy": 1.0,
          "op_accuracy": 1.0,
          "arg_accuracy": 1.0,
          "program_exact": 1.0
        },
        "len12": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 0.9375,
          "executor_target_mass": 0.9364005327224731,
          "init_accuracy": 1.0,
          "op_accuracy": 0.9947916666666666,
          "arg_accuracy": 1.0,
          "program_exact": 0.9375
        },
        "len16": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 0.921875,
          "executor_target_mass": 0.9153636693954468,
          "init_accuracy": 1.0,
          "op_accuracy": 0.99560546875,
          "arg_accuracy": 0.99951171875,
          "program_exact": 0.921875
        },
        "len24": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 0.875,
          "executor_target_mass": 0.8271066546440125,
          "init_accuracy": 1.0,
          "op_accuracy": 0.9947916666666666,
          "arg_accuracy": 1.0,
          "program_exact": 0.875
        }
      }
    }
  }
}