{
  "mode": "executor",
  "args": {
    "mode": "executor",
    "seed": 21,
    "modulus": 31,
    "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,
    "train_steps": 1000,
    "batch_size": 512,
    "eval_batch_size": 512,
    "eval_examples": 1024,
    "lr": 0.0005,
    "weight_decay": 0.01,
    "max_grad_norm": 1.0,
    "log_every": 50,
    "eval_every": 250,
    "output_dir": "runs_latent_executor/pilot_executor_mod31"
  },
  "train": [
    {
      "step": 1,
      "loss": 3.6087188720703125,
      "pair_acc": 0.0,
      "grad_norm": 0.5811763405799866,
      "gate_entropy": 10.999177098274231,
      "mem_rms": 0.13885721564292908
    },
    {
      "step": 50,
      "loss": 2.821141004562378,
      "pair_acc": 0.01953125,
      "grad_norm": 0.46821826696395874,
      "gate_entropy": 9.65117871761322,
      "mem_rms": 0.531837522983551
    },
    {
      "step": 100,
      "loss": 2.7674903869628906,
      "pair_acc": 0.017578125,
      "grad_norm": 0.4627982974052429,
      "gate_entropy": 8.767057657241821,
      "mem_rms": 0.5159477591514587
    },
    {
      "step": 150,
      "loss": 2.6937761306762695,
      "pair_acc": 0.015625,
      "grad_norm": 0.4883667826652527,
      "gate_entropy": 3.9208683520555496,
      "mem_rms": 0.6013717651367188
    },
    {
      "step": 200,
      "loss": 2.562485694885254,
      "pair_acc": 0.017578125,
      "grad_norm": 0.44592782855033875,
      "gate_entropy": 0.8430413641035557,
      "mem_rms": 0.9574517607688904
    },
    {
      "step": 250,
      "loss": 2.5297200679779053,
      "pair_acc": 0.017578125,
      "grad_norm": 0.43804192543029785,
      "gate_entropy": 0.6040909551084042,
      "mem_rms": 0.9890725612640381
    },
    {
      "step": 300,
      "loss": 2.531442165374756,
      "pair_acc": 0.009765625,
      "grad_norm": 0.43692418932914734,
      "gate_entropy": 0.40271352604031563,
      "mem_rms": 1.0519611835479736
    },
    {
      "step": 350,
      "loss": 2.4054431915283203,
      "pair_acc": 0.01171875,
      "grad_norm": 0.40071582794189453,
      "gate_entropy": 0.26328720431774855,
      "mem_rms": 1.030394196510315
    },
    {
      "step": 400,
      "loss": 2.42744779586792,
      "pair_acc": 0.025390625,
      "grad_norm": 0.3955957293510437,
      "gate_entropy": 0.2711191773414612,
      "mem_rms": 0.9590599536895752
    },
    {
      "step": 450,
      "loss": 2.4264488220214844,
      "pair_acc": 0.029296875,
      "grad_norm": 0.343288391828537,
      "gate_entropy": 0.33434971794486046,
      "mem_rms": 0.9640104174613953
    },
    {
      "step": 500,
      "loss": 2.432781219482422,
      "pair_acc": 0.01171875,
      "grad_norm": 0.4112797975540161,
      "gate_entropy": 0.43623500131070614,
      "mem_rms": 0.8641354441642761
    }
  ],
  "eval": [
    {
      "step": 250,
      "rows": [
        {
          "model": "executor",
          "length": 4,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.01171875,
          "register_accuracy": 0.115234375
        },
        {
          "model": "executor",
          "length": 4,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0107421875,
          "register_accuracy": 0.11376953125
        },
        {
          "model": "executor",
          "length": 4,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0068359375,
          "register_accuracy": 0.11328125
        },
        {
          "model": "executor",
          "length": 4,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.009765625,
          "register_accuracy": 0.11572265625
        },
        {
          "model": "executor",
          "length": 4,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.009765625,
          "register_accuracy": 0.11572265625
        },
        {
          "model": "executor",
          "length": 4,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.009765625,
          "register_accuracy": 0.11572265625
        },
        {
          "model": "executor",
          "length": 4,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.009765625,
          "register_accuracy": 0.11572265625
        },
        {
          "model": "executor",
          "length": 4,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.009765625,
          "register_accuracy": 0.11572265625
        },
        {
          "model": "executor",
          "length": 8,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.044921875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.04052734375
        },
        {
          "model": "executor",
          "length": 8,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0439453125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.04345703125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.0439453125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.0439453125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.0439453125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.0439453125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03564453125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.037109375
        },
        {
          "model": "executor",
          "length": 12,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03759765625
        },
        {
          "model": "executor",
          "length": 12,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.0302734375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.029296875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.0302734375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.033203125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.029296875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03173828125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0302734375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0302734375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.029296875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03369140625
        },
        {
          "model": "executor",
          "length": 24,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0341796875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03271484375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03173828125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03759765625
        },
        {
          "model": "executor",
          "length": 24,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0341796875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.0341796875
        }
      ]
    },
    {
      "step": 500,
      "rows": [
        {
          "model": "executor",
          "length": 4,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.01171875,
          "register_accuracy": 0.09814453125
        },
        {
          "model": "executor",
          "length": 4,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0107421875,
          "register_accuracy": 0.1064453125
        },
        {
          "model": "executor",
          "length": 4,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.01171875,
          "register_accuracy": 0.111328125
        },
        {
          "model": "executor",
          "length": 4,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0126953125,
          "register_accuracy": 0.115234375
        },
        {
          "model": "executor",
          "length": 4,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0126953125,
          "register_accuracy": 0.115234375
        },
        {
          "model": "executor",
          "length": 4,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0126953125,
          "register_accuracy": 0.115234375
        },
        {
          "model": "executor",
          "length": 4,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0126953125,
          "register_accuracy": 0.115234375
        },
        {
          "model": "executor",
          "length": 4,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0126953125,
          "register_accuracy": 0.115234375
        },
        {
          "model": "executor",
          "length": 8,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.00390625,
          "register_accuracy": 0.04345703125
        },
        {
          "model": "executor",
          "length": 8,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.00390625,
          "register_accuracy": 0.0390625
        },
        {
          "model": "executor",
          "length": 8,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.0048828125,
          "register_accuracy": 0.04541015625
        },
        {
          "model": "executor",
          "length": 8,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.04443359375
        },
        {
          "model": "executor",
          "length": 8,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.04248046875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.04248046875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.04248046875
        },
        {
          "model": "executor",
          "length": 8,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.04248046875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03173828125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.0322265625
        },
        {
          "model": "executor",
          "length": 12,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.033203125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03955078125
        },
        {
          "model": "executor",
          "length": 12,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 12,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.0263671875
        },
        {
          "model": "executor",
          "length": 16,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.03173828125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03271484375
        },
        {
          "model": "executor",
          "length": 16,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.03125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.02978515625
        },
        {
          "model": "executor",
          "length": 16,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.0,
          "register_accuracy": 0.0283203125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.02783203125
        },
        {
          "model": "executor",
          "length": 16,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.02783203125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 0,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.0283203125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 1,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.02685546875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 2,
          "n": 1024,
          "pair_accuracy": 0.00390625,
          "register_accuracy": 0.03564453125
        },
        {
          "model": "executor",
          "length": 24,
          "k": 4,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.0341796875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 8,
          "n": 1024,
          "pair_accuracy": 0.0009765625,
          "register_accuracy": 0.03466796875
        },
        {
          "model": "executor",
          "length": 24,
          "k": 12,
          "n": 1024,
          "pair_accuracy": 0.001953125,
          "register_accuracy": 0.02978515625
        },
        {
          "model": "executor",
          "length": 24,
          "k": 16,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.02734375
        },
        {
          "model": "executor",
          "length": 24,
          "k": 24,
          "n": 1024,
          "pair_accuracy": 0.0029296875,
          "register_accuracy": 0.029296875
        }
      ]
    }
  ]
}