simbert_distill_4L192H / config.json
WangZeJun's picture
Upload config.json
317ca26
raw
history blame contribute delete
332 Bytes
{
"attention_probs_dropout_prob": 0.1,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 192,
"initializer_range": 0.02,
"intermediate_size": 768,
"max_position_embeddings": 512,
"num_attention_heads": 12,
"num_hidden_layers": 4,
"pad_token_id": 0,
"type_vocab_size": 2,
"vocab_size": 13685
}