2019-10-17 19:12:53 +00:00
|
|
|
{
|
2020-01-31 23:00:24 +00:00
|
|
|
"architectures": [
|
|
|
|
"RobertaForMaskedLM"
|
|
|
|
],
|
2019-10-17 19:12:53 +00:00
|
|
|
"attention_probs_dropout_prob": 0.1,
|
2020-04-24 15:58:11 +00:00
|
|
|
"bos_token_id": 0,
|
|
|
|
"eos_token_id": 2,
|
2019-10-17 19:12:53 +00:00
|
|
|
"hidden_act": "gelu",
|
|
|
|
"hidden_dropout_prob": 0.1,
|
|
|
|
"hidden_size": 768,
|
|
|
|
"initializer_range": 0.02,
|
|
|
|
"intermediate_size": 3072,
|
|
|
|
"layer_norm_eps": 1e-05,
|
|
|
|
"max_position_embeddings": 514,
|
2020-04-24 15:58:11 +00:00
|
|
|
"model_type": "roberta",
|
2019-10-17 19:12:53 +00:00
|
|
|
"num_attention_heads": 12,
|
|
|
|
"num_hidden_layers": 6,
|
2020-04-24 15:58:11 +00:00
|
|
|
"pad_token_id": 1,
|
2019-10-17 19:12:53 +00:00
|
|
|
"type_vocab_size": 1,
|
|
|
|
"vocab_size": 50265
|
|
|
|
}
|