{
  "conv_layers": [
    [
      512,
      10,
      5
    ],
    [
      512,
      3,
      2
    ],
    [
      512,
      3,
      2
    ],
    [
      512,
      3,
      2
    ],
    [
      512,
      3,
      2
    ],
    [
      512,
      2,
      2
    ],
    [
      512,
      2,
      2
    ]
  ],
  "num_heads": [
    2,
    5,
    4,
    5,
    3,
    10,
    8,
    4,
    1,
    3,
    9,
    5,
    1,
    4,
    5,
    8,
    5,
    11,
    2,
    3,
    0,
    0,
    0,
    0
  ],
  "ffn_dims": [
    506,
    326,
    309,
    362,
    554,
    1112,
    114,
    148,
    266,
    481,
    991,
    1713,
    106,
    132,
    218,
    300,
    649,
    2117,
    64,
    159,
    132,
    202,
    270,
    432
  ],
  "model_dim": 1024,
  "head_dim": 64,
  "pos_conv_kernel": 128,
  "pos_conv_groups": 16,
  "vocab_size": 9812,
  "ffn_dropout": 0.1,
  "attn_dropout": 0.0,
  "layerdrop": 0.05,
  "feature_grad_scale": 0.1,
  "prune_conv": false,
  "prune_heads": false,
  "prune_interm": false,
  "prune_attn_layer": false,
  "prune_ffn_layer": false
}