{
  "mode": "categorical",
  "args": {
    "mode": "categorical",
    "seed": 22,
    "modulus": 31,
    "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": 2048,
    "eval_batch_size": 1024,
    "eval_examples": 2048,
    "lr": 0.05,
    "weight_decay": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 50,
    "eval_every": 200,
    "output_dir": "runs_latent_executor/pilot_categorical_mod31"
  },
  "train": [
    {
      "step": 1,
      "loss": 2.435689926147461,
      "pair_acc": 0.0126953125,
      "grad_norm": 0.008454954251646996,
      "transition_entropy": 3.4339386224746704
    },
    {
      "step": 50,
      "loss": 1.0655531883239746,
      "pair_acc": 0.9501953125,
      "grad_norm": 0.012477499432861805,
      "transition_entropy": 2.9388628005981445
    },
    {
      "step": 100,
      "loss": 0.09551163017749786,
      "pair_acc": 1.0,
      "grad_norm": 0.0018619407201185822,
      "transition_entropy": 1.7284188456833363
    },
    {
      "step": 150,
      "loss": 0.0427529513835907,
      "pair_acc": 1.0,
      "grad_norm": 0.0008422914543189108,
      "transition_entropy": 1.6306307055056095
    },
    {
      "step": 200,
      "loss": 0.026874398812651634,
      "pair_acc": 1.0,
      "grad_norm": 0.0005369526334106922,
      "transition_entropy": 1.5754426270723343
    },
    {
      "step": 250,
      "loss": 0.018922697752714157,
      "pair_acc": 1.0,
      "grad_norm": 0.0003791946801356971,
      "transition_entropy": 1.558772323653102
    },
    {
      "step": 300,
      "loss": 0.014388136565685272,
      "pair_acc": 1.0,
      "grad_norm": 0.00028890042449347675,
      "transition_entropy": 1.5254616420716047
    },
    {
      "step": 350,
      "loss": 0.011192583478987217,
      "pair_acc": 1.0,
      "grad_norm": 0.0002252909034723416,
      "transition_entropy": 1.5358409015461802
    },
    {
      "step": 400,
      "loss": 0.009154919534921646,
      "pair_acc": 1.0,
      "grad_norm": 0.0001839060860220343,
      "transition_entropy": 1.5252246093004942
    }
  ],
  "eval": [
    {
      "step": 200,
      "rows": [
        {
          "model": "executor",
          "length": 4,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.00732421875,
          "register_accuracy": 0.08935546875
        },
        {
          "model": "executor",
          "length": 4,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00634765625,
          "register_accuracy": 0.145263671875
        },
        {
          "model": "executor",
          "length": 4,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0166015625,
          "register_accuracy": 0.260498046875
        },
        {
          "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": 0.00146484375,
          "register_accuracy": 0.03759765625
        },
        {
          "model": "executor",
          "length": 8,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.0419921875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.045654296875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.00439453125,
          "register_accuracy": 0.08447265625
        },
        {
          "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": 0.03076171875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.03125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.037353515625
        },
        {
          "model": "executor",
          "length": 12,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.036376953125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.00390625,
          "register_accuracy": 0.08447265625
        },
        {
          "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": 0.001953125,
          "register_accuracy": 0.034912109375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03515625
        },
        {
          "model": "executor",
          "length": 16,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.02734375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03515625
        },
        {
          "model": "executor",
          "length": 16,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.034912109375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.0888671875
        },
        {
          "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": 0.0009765625,
          "register_accuracy": 0.032470703125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.034912109375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.032470703125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.029052734375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.032470703125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.03271484375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.038818359375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        }
      ]
    },
    {
      "step": 400,
      "rows": [
        {
          "model": "executor",
          "length": 4,
          "k": 0,
          "n": 2048,
          "pair_accuracy": 0.00634765625,
          "register_accuracy": 0.085205078125
        },
        {
          "model": "executor",
          "length": 4,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.009765625,
          "register_accuracy": 0.141357421875
        },
        {
          "model": "executor",
          "length": 4,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.015625,
          "register_accuracy": 0.254638671875
        },
        {
          "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": 0.034423828125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00244140625,
          "register_accuracy": 0.039794921875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.04638671875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0068359375,
          "register_accuracy": 0.092529296875
        },
        {
          "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": 0.0009765625,
          "register_accuracy": 0.031982421875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.029052734375
        },
        {
          "model": "executor",
          "length": 12,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.029296875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.035400390625
        },
        {
          "model": "executor",
          "length": 12,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.00244140625,
          "register_accuracy": 0.0771484375
        },
        {
          "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": 0.037841796875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.00146484375,
          "register_accuracy": 0.039306640625
        },
        {
          "model": "executor",
          "length": 16,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.0380859375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.032958984375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.0390625
        },
        {
          "model": "executor",
          "length": 16,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 0.0048828125,
          "register_accuracy": 0.08740234375
        },
        {
          "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": 0.0009765625,
          "register_accuracy": 0.033935546875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 1,
          "n": 2048,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.034423828125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 2,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.033447265625
        },
        {
          "model": "executor",
          "length": 24,
          "k": 4,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.031005859375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 8,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03369140625
        },
        {
          "model": "executor",
          "length": 24,
          "k": 12,
          "n": 2048,
          "pair_accuracy": 0.00048828125,
          "register_accuracy": 0.027099609375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 16,
          "n": 2048,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.040283203125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 24,
          "n": 2048,
          "pair_accuracy": 1.0,
          "register_accuracy": 1.0
        }
      ]
    }
  ]
}