{
  "d_model": 512,
  "depth": 12,
  "expansion_factor": 1.5,
  "ff_mult": 4,
  "hidden_dropout_prob": 0.1,
  "hidden_size": 512,
  "initializer_range": 0.02,
  "model_type": "mingru",
  "n_layer": 12,
  "pad_vocab_size_multiple": 8,
  "transformers_version": "4.44.2",
  "vocab_size": 30522
}