from transformers import BertConfig, AutoConfig, BertForPreTraining
bert_config = {'attention_probs_dropout_prob': 0.1,
'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': 2,
'vocab_size': 8002}
config = BertConfig.from_dict(bert_config)
print(config)
config = BertConfig.from_pretrained("bert-base-cased")
print(config)
config = AutoConfig.from_pretrained("bert-base-cased")
print(config)
model = BertForPreTraining.from_pretrained('bert-base-cased')
config = model.config
print(config)