{
  "mode": "categorical",
  "args": {
    "mode": "categorical",
    "seed": 23,
    "modulus": 97,
    "op_family": "const",
    "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",
    "dim": 128,
    "heads": 4,
    "compiler_layers": 1,
    "num_bases": 16,
    "rank": 16,
    "mem_dim": 48,
    "dropout": 0.05,
    "disable_memory": false,
    "disable_dynamic": false,
    "transition_temperature": 1.0,
    "train_steps": 1000,
    "batch_size": 4096,
    "eval_batch_size": 512,
    "eval_examples": 2048,
    "lr": 0.05,
    "weight_decay": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 50,
    "eval_every": 250,
    "output_dir": "runs_latent_executor/categorical_mod97"
  },
  "train": [
    {
      "step": 1,
      "loss": 3.2330331802368164,
      "pair_acc": 0.002197265625,
      "grad_norm": 0.005031558219343424,
      "transition_entropy": 4.5746612548828125
    },
    {
      "step": 50,
      "loss": 2.7870101928710938,
      "pair_acc": 0.2001953125,
      "grad_norm": 0.005093299318104982,
      "transition_entropy": 4.544336915016174
    },
    {
      "step": 100,
      "loss": 1.9014382362365723,
      "pair_acc": 0.564697265625,
      "grad_norm": 0.006888106931000948,
      "transition_entropy": 4.272156715393066
    },
    {
      "step": 150,
      "loss": 0.3860275149345398,
      "pair_acc": 1.0,
      "grad_norm": 0.0032750386744737625,
      "transition_entropy": 2.8756475299596786
    },
    {
      "step": 200,
      "loss": 0.12852060794830322,
      "pair_acc": 1.0,
      "grad_norm": 0.001182078616693616,
      "transition_entropy": 2.4030509665608406
    },
    {
      "step": 250,
      "loss": 0.07450802624225616,
      "pair_acc": 1.0,
      "grad_norm": 0.0006960609462112188,
      "transition_entropy": 2.2779110819101334
    }
  ],
  "eval": [
    {
      "step": 250,
      "rows": [
        {
          "model": "executor",
          "length": 4,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.07421875
        },
        {
          "model": "executor",
          "length": 4,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.137451171875
        },
        {
          "model": "executor",
          "length": 4,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.005859375,
          "register_accuracy": 0.259521484375
        },
        {
          "model": "executor",
          "length": 4,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 4,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 4,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 4,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 4,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 8,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.014404296875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.017333984375
        },
        {
          "model": "executor",
          "length": 8,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.02734375
        },
        {
          "model": "executor",
          "length": 8,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.0703125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 8,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 8,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 8,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 12,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.012451171875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.011474609375
        },
        {
          "model": "executor",
          "length": 12,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.010009765625
        },
        {
          "model": "executor",
          "length": 12,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.01318359375
        },
        {
          "model": "executor",
          "length": 12,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.066650390625
        },
        {
          "model": "executor",
          "length": 12,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 12,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 12,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 16,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.010986328125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.010986328125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.01171875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.008544921875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0126953125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 0.00244140625,
          "register_accuracy": 0.076171875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 16,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        },
        {
          "model": "executor",
          "length": 24,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.010986328125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0107421875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.01123046875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.011474609375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.00927734375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0126953125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.01318359375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        }
      ]
    }
  ]
}