mdeberta-v3-base / generator_config.json
Pengcheng He
Add generator
a048466
raw
history blame contribute delete
575 Bytes
{
"model_type": "deberta-v2",
"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,
"max_relative_positions": -1,
"num_attention_heads": 12,
"num_hidden_layers": 6,
"pos_att_type": "c2p|p2c",
"position_biased_input": false,
"position_buckets": 256,
"norm_rel_ebd": "layer_norm",
"relative_attention": true,
"share_att_key": true,
"type_vocab_size": 0,
"layer_norm_eps": 1e-7,
"vocab_size": 251000
}