{
  "mode": "static",
  "args": {
    "mode": "static",
    "task": "belief_line",
    "seed": 23,
    "modulus": 11,
    "op_family": "full",
    "train_min_len": 1,
    "train_max_len": 6,
    "eval_lengths": "3,6,9,12",
    "eval_k": "0,1,2,3,6,9,12",
    "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": 25,
    "eval_every": 100,
    "output_dir": "experiments/joint_register_executor/runs/control_belief_static_mod11",
    "checkpoint_dir": "large_artifacts/joint_register_executor/checkpoints/control_belief_static_mod11"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 4.81330680847168,
      "pair_acc": 0.08984375,
      "grad_norm": 0.20231597125530243
    },
    {
      "step": 25,
      "loss": 4.780508518218994,
      "pair_acc": 0.146484375,
      "grad_norm": 0.09643480181694031
    },
    {
      "step": 50,
      "loss": 4.695324420928955,
      "pair_acc": 0.2578125,
      "grad_norm": 0.1263684332370758
    },
    {
      "step": 75,
      "loss": 4.615285873413086,
      "pair_acc": 0.244140625,
      "grad_norm": 0.125777930021286
    },
    {
      "step": 100,
      "loss": 4.513310432434082,
      "pair_acc": 0.23828125,
      "grad_norm": 0.09511519223451614
    },
    {
      "step": 125,
      "loss": 4.406520843505859,
      "pair_acc": 0.287109375,
      "grad_norm": 0.11476808041334152
    },
    {
      "step": 150,
      "loss": 4.50686502456665,
      "pair_acc": 0.25390625,
      "grad_norm": 0.21332748234272003
    },
    {
      "step": 175,
      "loss": 4.409083366394043,
      "pair_acc": 0.267578125,
      "grad_norm": 0.1270049810409546
    },
    {
      "step": 200,
      "loss": 4.363996505737305,
      "pair_acc": 0.296875,
      "grad_norm": 0.1544819474220276
    },
    {
      "step": 225,
      "loss": 4.403125762939453,
      "pair_acc": 0.271484375,
      "grad_norm": 0.15729042887687683
    },
    {
      "step": 250,
      "loss": 4.374475002288818,
      "pair_acc": 0.294921875,
      "grad_norm": 0.15105579793453217
    },
    {
      "step": 275,
      "loss": 4.3181633949279785,
      "pair_acc": 0.306640625,
      "grad_norm": 0.19197875261306763
    },
    {
      "step": 300,
      "loss": 4.289544105529785,
      "pair_acc": 0.2890625,
      "grad_norm": 0.1584940403699875
    }
  ],
  "eval": [
    {
      "step": 100,
      "rows": [
        {
          "model": "static",
          "length": 3,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.17578125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.1218254560546875,
          "target_nll": 4.6795709140625,
          "top1_on_support": 0.17578125
        },
        {
          "model": "static",
          "length": 6,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.10546875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0962012109375,
          "target_nll": 4.782914638671875,
          "top1_on_support": 0.10546875
        },
        {
          "model": "static",
          "length": 9,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.0986328125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.09100092578125,
          "target_nll": 4.80517458984375,
          "top1_on_support": 0.0986328125
        },
        {
          "model": "static",
          "length": 12,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.076171875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0911910576171875,
          "target_nll": 4.802636146484375,
          "top1_on_support": 0.076171875
        }
      ]
    },
    {
      "step": 200,
      "rows": [
        {
          "model": "static",
          "length": 3,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.2421875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.166449501953125,
          "target_nll": 4.55965066015625,
          "top1_on_support": 0.2421875
        },
        {
          "model": "static",
          "length": 6,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.1123046875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0959849365234375,
          "target_nll": 4.783015250976563,
          "top1_on_support": 0.1123046875
        },
        {
          "model": "static",
          "length": 9,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.087890625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0930548427734375,
          "target_nll": 4.793198109375,
          "top1_on_support": 0.087890625
        },
        {
          "model": "static",
          "length": 12,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.1015625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.09187921875,
          "target_nll": 4.7991352080078125,
          "top1_on_support": 0.1015625
        }
      ]
    },
    {
      "step": 300,
      "rows": [
        {
          "model": "static",
          "length": 3,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.2529296875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.1844290498046875,
          "target_nll": 4.500460385742188,
          "top1_on_support": 0.2529296875
        },
        {
          "model": "static",
          "length": 6,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.1103515625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.101009751953125,
          "target_nll": 4.765430926757812,
          "top1_on_support": 0.1103515625
        },
        {
          "model": "static",
          "length": 9,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.0986328125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0914890693359375,
          "target_nll": 4.797176838867188,
          "top1_on_support": 0.0986328125
        },
        {
          "model": "static",
          "length": 12,
          "k": -1,
          "n": 1024,
          "pair_accuracy": 0.1005859375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0910594765625,
          "target_nll": 4.801716328125,
          "top1_on_support": 0.1005859375
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/joint_register_executor/checkpoints/control_belief_static_mod11/checkpoint_step00100.pt",
    "large_artifacts/joint_register_executor/checkpoints/control_belief_static_mod11/checkpoint_step00200.pt",
    "large_artifacts/joint_register_executor/checkpoints/control_belief_static_mod11/checkpoint_step00300.pt"
  ]
}