bert4torch_config.json

{
  "attention_probs_dropout_prob": 0.1, 
  "directionality": "bidi", 
  "hidden_act": "gelu", 
  "hidden_dropout_prob": 0.1, 
  "hidden_size": 768, 
  "initializer_range": 0.02, 
  "intermediate_size": 3072, 
  "max_position_embeddings": 512, 
  "num_attention_heads": 12, 
  "num_hidden_layers": 12, 
  "type_vocab_size": 0, 
  "vocab_size": 21128,
  "application": "lm"
}