bert4torch_config.json

{
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 768,
  "initializer_range": 0.02,
  "intermediate_size": 2048,
  "num_attention_heads": 12,
  "attention_head_size": 64,
  "num_hidden_layers": 12,
  "vocab_size": 50000,
  "relative_attention_num_buckets": 32,
  "attention_scale":  false,
  "is_dropout": true,
  "model": "mt5.1.1",
  "segment_vocab_size": 0
}