2019-12-13 23:30:34 +08:00
|
|
|
{
|
2020-04-24 23:58:05 +08:00
|
|
|
"activation": "gelu",
|
2020-02-01 07:00:24 +08:00
|
|
|
"architectures": [
|
|
|
|
|
"DistilBertForSequenceClassification"
|
|
|
|
|
],
|
2019-12-13 23:30:34 +08:00
|
|
|
"attention_dropout": 0.1,
|
|
|
|
|
"dim": 768,
|
|
|
|
|
"dropout": 0.1,
|
|
|
|
|
"finetuning_task": "sst-2",
|
|
|
|
|
"hidden_dim": 3072,
|
2020-04-24 23:58:05 +08:00
|
|
|
"id2label": {
|
|
|
|
|
"0": "NEGATIVE",
|
|
|
|
|
"1": "POSITIVE"
|
|
|
|
|
},
|
2019-12-13 23:30:34 +08:00
|
|
|
"initializer_range": 0.02,
|
2020-04-24 23:58:05 +08:00
|
|
|
"label2id": {
|
|
|
|
|
"NEGATIVE": 0,
|
|
|
|
|
"POSITIVE": 1
|
|
|
|
|
},
|
2019-12-13 23:30:34 +08:00
|
|
|
"max_position_embeddings": 512,
|
2020-04-24 23:58:05 +08:00
|
|
|
"model_type": "distilbert",
|
2019-12-13 23:30:34 +08:00
|
|
|
"n_heads": 12,
|
|
|
|
|
"n_layers": 6,
|
|
|
|
|
"output_past": true,
|
2020-04-24 23:58:05 +08:00
|
|
|
"pad_token_id": 0,
|
2019-12-13 23:30:34 +08:00
|
|
|
"qa_dropout": 0.1,
|
|
|
|
|
"seq_classif_dropout": 0.2,
|
|
|
|
|
"sinusoidal_pos_embds": false,
|
|
|
|
|
"tie_weights_": true,
|
2020-04-24 23:58:05 +08:00
|
|
|
"vocab_size": 30522
|
2019-12-13 23:30:34 +08:00
|
|
|
}
|
