{
  "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": 12,
    "train_min_len": 1,
    "train_max_len": 3,
    "answer_train_min_len": 1,
    "answer_train_max_len": 3,
    "eval_lengths": "3,6,12",
    "train_template_mode": "standard",
    "answer_train_template_mode": "standard",
    "eval_template_modes": "standard",
    "train_size": 1024,
    "answer_train_size": 128,
    "eval_size": 128,
    "feature_batch_size": 8,
    "train_batch_size": 128,
    "eval_batch_size": 128,
    "bootstrap_steps": 1000,
    "answer_steps": 0,
    "eval_every": 500,
    "stage_eval_every": 500,
    "lr": 0.001,
    "answer_lr_scale": 0.1,
    "weight_decay": 0.01,
    "grad_clip": 1.0,
    "head_width": 512,
    "trace_loss_weight": 1.0,
    "init_trace_loss_weight": 4.0,
    "op_trace_loss_weight": 1.0,
    "arg_trace_loss_weight": 1.0,
    "attention_loss_weight": 1.0,
    "executor_loss_weight": 1.0,
    "variants": "compiler_trace",
    "max_length": 512,
    "feature_seq_len": 192,
    "seed": 7,
    "output_dir": "experiments/qwen_span_free_compiler/runs/pilot_qwen35_generated_query_attention",
    "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": {
    "bootstrap_train_size": 1024,
    "answer_train_size": 128,
    "eval_size": 128,
    "bootstrap_train_lengths": [
      1,
      3
    ],
    "answer_train_lengths": [
      1,
      3
    ],
    "eval_lengths": [
      3,
      6,
      12
    ],
    "train_template_mode": "standard",
    "answer_train_template_mode": "standard",
    "eval_template_modes": [
      "standard"
    ],
    "feature_seq_len": 192
  },
  "variants": {
    "compiler_trace": {
      "variant": "compiler_trace",
      "train_seconds": 80.39535641670227,
      "checkpoints": [
        "large_artifacts/qwen_span_free_compiler/checkpoints/pilot_qwen35_generated_query_attention/compiler_trace/checkpoint_final.pt"
      ],
      "train_log": [
        {
          "stage": "trace_bootstrap",
          "local_step": 1,
          "step": 1,
          "loss": 41.0858154296875,
          "trace_loss_active": true,
          "lr_scale": 1.0,
          "executor_loss": 4.570854187011719,
          "init_loss": 4.555853366851807,
          "op_loss": 1.1046651601791382,
          "arg_loss": 4.645595550537109,
          "init_attn_loss": 4.027915000915527,
          "op_attn_loss": 4.356146335601807,
          "arg_attn_loss": 4.157222747802734,
          "standard_len3_direct_accuracy": NaN,
          "standard_len3_direct_target_mass": NaN,
          "standard_len3_executor_accuracy": 0.0078125,
          "standard_len3_executor_target_mass": 0.010313953272998333,
          "standard_len3_init_accuracy": 0.0,
          "standard_len3_op_accuracy": 0.3151041666666667,
          "standard_len3_arg_accuracy": 0.041666666666666664,
          "standard_len3_program_exact": 0.0,
          "standard_len6_direct_accuracy": NaN,
          "standard_len6_direct_target_mass": NaN,
          "standard_len6_executor_accuracy": 0.015625,
          "standard_len6_executor_target_mass": 0.010307900607585907,
          "standard_len6_init_accuracy": 0.0,
          "standard_len6_op_accuracy": 0.3385416666666667,
          "standard_len6_arg_accuracy": 0.052083333333333336,
          "standard_len6_program_exact": 0.0,
          "standard_len12_direct_accuracy": NaN,
          "standard_len12_direct_target_mass": NaN,
          "standard_len12_executor_accuracy": 0.0,
          "standard_len12_executor_target_mass": 0.010304208844900131,
          "standard_len12_init_accuracy": 0.015625,
          "standard_len12_op_accuracy": 0.3274739583333333,
          "standard_len12_arg_accuracy": 0.045572916666666664,
          "standard_len12_program_exact": 0.0
        },
        {
          "stage": "trace_bootstrap",
          "local_step": 500,
          "step": 500,
          "loss": 6.543615745613351e-05,
          "trace_loss_active": true,
          "lr_scale": 1.0,
          "executor_loss": 9.374753062729724e-06,
          "init_loss": 6.969988135097083e-06,
          "op_loss": 1.6834883709293536e-08,
          "arg_loss": 1.1579631973290816e-06,
          "init_attn_loss": 4.114514013053849e-06,
          "op_attn_loss": 1.7062274082491058e-06,
          "arg_attn_loss": 2.1185909645282663e-05,
          "standard_len3_direct_accuracy": NaN,
          "standard_len3_direct_target_mass": NaN,
          "standard_len3_executor_accuracy": 0.9921875,
          "standard_len3_executor_target_mass": 0.9921561479568481,
          "standard_len3_init_accuracy": 1.0,
          "standard_len3_op_accuracy": 1.0,
          "standard_len3_arg_accuracy": 0.9973958333333334,
          "standard_len3_program_exact": 0.9921875,
          "standard_len6_direct_accuracy": NaN,
          "standard_len6_direct_target_mass": NaN,
          "standard_len6_executor_accuracy": 0.0078125,
          "standard_len6_executor_target_mass": 0.010552351363003254,
          "standard_len6_init_accuracy": 1.0,
          "standard_len6_op_accuracy": 0.59765625,
          "standard_len6_arg_accuracy": 0.3567708333333333,
          "standard_len6_program_exact": 0.0,
          "standard_len12_direct_accuracy": NaN,
          "standard_len12_direct_target_mass": NaN,
          "standard_len12_executor_accuracy": 0.0,
          "standard_len12_executor_target_mass": 0.011335321702063084,
          "standard_len12_init_accuracy": 1.0,
          "standard_len12_op_accuracy": 0.494140625,
          "standard_len12_arg_accuracy": 0.20377604166666666,
          "standard_len12_program_exact": 0.0
        },
        {
          "stage": "trace_bootstrap",
          "local_step": 1000,
          "step": 1000,
          "loss": 3.401468711672351e-05,
          "trace_loss_active": true,
          "lr_scale": 1.0,
          "executor_loss": 4.67897780254134e-06,
          "init_loss": 3.544606443028897e-06,
          "op_loss": 1.2427218365473891e-08,
          "arg_loss": 5.481778657667746e-07,
          "init_attn_loss": 1.5329463849411695e-06,
          "op_attn_loss": 6.08472760177392e-07,
          "arg_attn_loss": 1.2455258001864422e-05,
          "standard_len3_direct_accuracy": NaN,
          "standard_len3_direct_target_mass": NaN,
          "standard_len3_executor_accuracy": 0.9921875,
          "standard_len3_executor_target_mass": 0.9921770095825195,
          "standard_len3_init_accuracy": 1.0,
          "standard_len3_op_accuracy": 1.0,
          "standard_len3_arg_accuracy": 0.9973958333333334,
          "standard_len3_program_exact": 0.9921875,
          "standard_len6_direct_accuracy": NaN,
          "standard_len6_direct_target_mass": NaN,
          "standard_len6_executor_accuracy": 0.0,
          "standard_len6_executor_target_mass": 0.009623616002500057,
          "standard_len6_init_accuracy": 1.0,
          "standard_len6_op_accuracy": 0.5859375,
          "standard_len6_arg_accuracy": 0.3528645833333333,
          "standard_len6_program_exact": 0.0,
          "standard_len12_direct_accuracy": NaN,
          "standard_len12_direct_target_mass": NaN,
          "standard_len12_executor_accuracy": 0.015625,
          "standard_len12_executor_target_mass": 0.015305858105421066,
          "standard_len12_init_accuracy": 1.0,
          "standard_len12_op_accuracy": 0.4915364583333333,
          "standard_len12_arg_accuracy": 0.20442708333333334,
          "standard_len12_program_exact": 0.0
        }
      ],
      "final_metrics": {
        "standard_len3": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 0.9921875,
          "executor_target_mass": 0.9921770095825195,
          "init_accuracy": 1.0,
          "op_accuracy": 1.0,
          "arg_accuracy": 0.9973958333333334,
          "program_exact": 0.9921875
        },
        "standard_len6": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 0.0,
          "executor_target_mass": 0.009623616002500057,
          "init_accuracy": 1.0,
          "op_accuracy": 0.5859375,
          "arg_accuracy": 0.3528645833333333,
          "program_exact": 0.0
        },
        "standard_len12": {
          "variant": "compiler_trace",
          "n": 128.0,
          "direct_accuracy": NaN,
          "direct_target_mass": NaN,
          "executor_accuracy": 0.015625,
          "executor_target_mass": 0.015305858105421066,
          "init_accuracy": 1.0,
          "op_accuracy": 0.4915364583333333,
          "arg_accuracy": 0.20442708333333334,
          "program_exact": 0.0
        }
      }
    }
  }
}