{
  "mode": "marginal",
  "args": {
    "mode": "marginal",
    "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": 200,
    "batch_size": 512,
    "eval_batch_size": 512,
    "eval_examples": 1024,
    "lr": 0.03,
    "weight_decay": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 50,
    "eval_every": 200,
    "output_dir": "experiments/joint_register_executor/runs/control_belief_marginal_mod31",
    "checkpoint_dir": "large_artifacts/joint_register_executor/checkpoints/control_belief_marginal_mod31"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 6.867959976196289,
      "pair_acc": 0.0234375,
      "grad_norm": 0.0014550037449225783,
      "transition_entropy": 3.4339387118816376
    },
    {
      "step": 50,
      "loss": 6.873574733734131,
      "pair_acc": 0.03125,
      "grad_norm": 0.0015051118098199368,
      "transition_entropy": 3.4320830702781677
    },
    {
      "step": 100,
      "loss": 6.872885704040527,
      "pair_acc": 0.0390625,
      "grad_norm": 0.0015823865542188287,
      "transition_entropy": 3.432099789381027
    },
    {
      "step": 150,
      "loss": 6.872852325439453,
      "pair_acc": 0.03125,
      "grad_norm": 0.0013447689125314355,
      "transition_entropy": 3.431942969560623
    },
    {
      "step": 200,
      "loss": 6.8732805252075195,
      "pair_acc": 0.02734375,
      "grad_norm": 0.0015391980996355414,
      "transition_entropy": 3.43185493350029
    }
  ],
  "eval": [
    {
      "step": 200,
      "rows": [
        {
          "model": "marginal",
          "length": 4,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0439453125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322580703125,
          "target_nll": 6.8679738046875,
          "top1_on_support": 0.0439453125
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.041015625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032268009765625,
          "target_nll": 6.870618104492188,
          "top1_on_support": 0.041015625
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0361328125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032227796875,
          "target_nll": 6.873484373046875,
          "top1_on_support": 0.0361328125
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323034521484375,
          "target_nll": 6.872351646484375,
          "top1_on_support": 0.0390625
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323034521484375,
          "target_nll": 6.872351646484375,
          "top1_on_support": 0.0390625
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323034521484375,
          "target_nll": 6.872351646484375,
          "top1_on_support": 0.0390625
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323034521484375,
          "target_nll": 6.872351646484375,
          "top1_on_support": 0.0390625
        },
        {
          "model": "marginal",
          "length": 4,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323034521484375,
          "target_nll": 6.872351646484375,
          "top1_on_support": 0.0390625
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0263671875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322580703125,
          "target_nll": 6.8679738046875,
          "top1_on_support": 0.0263671875
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.03125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032255869140625,
          "target_nll": 6.871136427734375,
          "top1_on_support": 0.03125
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0322265625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032253255859375,
          "target_nll": 6.8729119296875,
          "top1_on_support": 0.0322265625
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0263671875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322652978515625,
          "target_nll": 6.87352156640625,
          "top1_on_support": 0.0263671875
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.03125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322688671875,
          "target_nll": 6.873841046875,
          "top1_on_support": 0.03125
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.03125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322688671875,
          "target_nll": 6.873841046875,
          "top1_on_support": 0.03125
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.03125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322688671875,
          "target_nll": 6.873841046875,
          "top1_on_support": 0.03125
        },
        {
          "model": "marginal",
          "length": 8,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.03125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322688671875,
          "target_nll": 6.873841046875,
          "top1_on_support": 0.03125
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0361328125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322580703125,
          "target_nll": 6.8679738046875,
          "top1_on_support": 0.0361328125
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0361328125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032214666015625,
          "target_nll": 6.872432708984375,
          "top1_on_support": 0.0361328125
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.03515625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032217484375,
          "target_nll": 6.8737521171875,
          "top1_on_support": 0.03515625
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.03515625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322640986328125,
          "target_nll": 6.873607158203125,
          "top1_on_support": 0.03515625
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.02734375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.03233576953125,
          "target_nll": 6.871635674804687,
          "top1_on_support": 0.02734375
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0380859375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322828125,
          "target_nll": 6.87327098828125,
          "top1_on_support": 0.0380859375
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0380859375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322828125,
          "target_nll": 6.87327098828125,
          "top1_on_support": 0.0380859375
        },
        {
          "model": "marginal",
          "length": 12,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0380859375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322828125,
          "target_nll": 6.87327098828125,
          "top1_on_support": 0.0380859375
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0341796875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322580703125,
          "target_nll": 6.8679738046875,
          "top1_on_support": 0.0341796875
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.037109375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.03229030078125,
          "target_nll": 6.870046853515625,
          "top1_on_support": 0.037109375
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.033203125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032282130859375,
          "target_nll": 6.8719215390625,
          "top1_on_support": 0.033203125
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0400390625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322546474609375,
          "target_nll": 6.873965263671875,
          "top1_on_support": 0.0400390625
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.033203125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323304873046875,
          "target_nll": 6.8721978662109375,
          "top1_on_support": 0.033203125
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0341796875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322388837890625,
          "target_nll": 6.874903440429687,
          "top1_on_support": 0.0341796875
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0302734375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032259357421875,
          "target_nll": 6.874144553710938,
          "top1_on_support": 0.0302734375
        },
        {
          "model": "marginal",
          "length": 16,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0302734375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032259357421875,
          "target_nll": 6.874144553710938,
          "top1_on_support": 0.0302734375
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.029296875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322580703125,
          "target_nll": 6.8679738046875,
          "top1_on_support": 0.029296875
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.029296875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032254666015625,
          "target_nll": 6.871184826171875,
          "top1_on_support": 0.029296875
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.03125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322569755859375,
          "target_nll": 6.872852326171875,
          "top1_on_support": 0.03125
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0302734375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0323175693359375,
          "target_nll": 6.872220516601563,
          "top1_on_support": 0.0302734375
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0234375,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032256404296875,
          "target_nll": 6.87417340234375,
          "top1_on_support": 0.0234375
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.029296875,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032253302734375,
          "target_nll": 6.87443733203125,
          "top1_on_support": 0.029296875
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0322265625,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.0322518251953125,
          "target_nll": 6.8741860390625,
          "top1_on_support": 0.0322265625
        },
        {
          "model": "marginal",
          "length": 24,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0361328125,
          "a_accuracy": NaN,
          "b_accuracy": NaN,
          "register_accuracy": NaN,
          "target_mass": 0.032290205078125,
          "target_nll": 6.873314857421875,
          "top1_on_support": 0.0361328125
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/joint_register_executor/checkpoints/control_belief_marginal_mod31/checkpoint_step00200.pt"
  ]
}