bert4torch_config.json

{
  "model": "gau_alpha",
  "hidden_act": "swish",
  "hidden_size": 768,
  "hidden_dropout_prob": 0.1,
  "attention_probs_dropout_prob": 0.1,
  "num_attention_heads": 1,
  "attention_key_size": 128,
  "intermediate_size": 1536,
  "num_hidden_layers": 24,
  "type_vocab_size": 2,
  "vocab_size": 12000
}