{
  "mode": "static",
  "args": {
    "mode": "static",
    "task": "belief_line",
    "seed": 23,
    "modulus": 31,
    "op_family": "full",
    "train_min_len": 1,
    "train_max_len": 8,
    "eval_lengths": "4,8,12,16,24",
    "eval_k": "0,1,2,4,8,12,16,24",
    "transition_temperature": 1.0,
    "dim": 128,
    "heads": 4,
    "compiler_layers": 2,
    "dropout": 0.05,
    "train_steps": 300,
    "batch_size": 512,
    "eval_batch_size": 512,
    "eval_examples": 1024,
    "lr": 0.001,
    "weight_decay": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 50,
    "eval_every": 300,
    "output_dir": "experiments/joint_register_executor/runs/control_belief_static_mod31",
    "checkpoint_dir": "large_artifacts/joint_register_executor/checkpoints/control_belief_static_mod31"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 6.884504318237305,
      "pair_acc": 0.033203125,
      "grad_norm": 0.10843706876039505
    },
    {
      "step": 50,
      "loss": 6.857891082763672,
      "pair_acc": 0.0546875,
      "grad_norm": 0.054251689463853836
    },
    {
      "step": 100,
      "loss": 6.700958728790283,
      "pair_acc": 0.099609375,
      "grad_norm": 0.12390845268964767
    },
    {
      "step": 150,
      "loss": 6.660524368286133,
      "pair_acc": 0.12109375,
      "grad_norm": 0.08785446733236313
    },
    {
      "step": 200,
      "loss": 6.55086088180542,
      "pair_acc": 0.1328125,
      "grad_norm": 0.10796110332012177
    },
    {
      "step": 250,
      "loss": 6.478620529174805,
      "pair_acc": 0.166015625,
      "grad_norm": 0.07105932384729385
    },
    {
      "step": 300,
      "loss": 6.468823432922363,
      "pair_acc": 0.162109375,
      "grad_norm": 0.14223001897335052
    }
  ],
  "eval": [
    {
      "step": 300,
      "rows": [
        {
          "model": "static",
          "length": 4,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.1083984375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0788499453125,
          "target_nll": 6.692636012695313,
          "top1_on_support": 0.1083984375
        },
        {
          "model": "static",
          "length": 8,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.0380859375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0349287255859375,
          "target_nll": 6.857664346679687,
          "top1_on_support": 0.0380859375
        },
        {
          "model": "static",
          "length": 12,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.0302734375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322529013671875,
          "target_nll": 6.87038803125,
          "top1_on_support": 0.0302734375
        },
        {
          "model": "static",
          "length": 16,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.025390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032264884765625,
          "target_nll": 6.869582176757812,
          "top1_on_support": 0.025390625
        },
        {
          "model": "static",
          "length": 24,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.033203125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322535615234375,
          "target_nll": 6.869007825195313,
          "top1_on_support": 0.033203125
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/joint_register_executor/checkpoints/control_belief_static_mod31/checkpoint_step00300.pt"
  ]
}