{
  "emb_size": 1024,
  "feedforward_size": 4096,
  "hidden_size": 1024,
  "hidden_act": "relu",
  "heads_num": 16,
  "layers_num": 16,
  "decoder_layers_num": 16,
  "max_seq_length": 1024,
  "dropout": 0.1,
  "data_processor": "gsg",
  "embedding": ["word", "sinusoidalpos"],
  "remove_embedding_layernorm": true,
  "tgt_embedding": ["word", "sinusoidalpos"],
  "share_embedding": true,
  "encoder": "transformer",
  "mask": "fully_visible",
  "layernorm_positioning": "pre",
  "decoder": "transformer",
  "target": ["lm"],
  "has_lmtarget_bias": true,
  "tie_weights": true
}