Upload config.json with huggingface_hub
Browse files- config.json +40 -0
config.json
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "microsoft/mdeberta-v3-base",
|
3 |
+
"architectures": [
|
4 |
+
"DebertaV2ForSequenceClassification"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"finetuning_task": "vnrte",
|
8 |
+
"hidden_act": "gelu",
|
9 |
+
"hidden_dropout_prob": 0.1,
|
10 |
+
"hidden_size": 768,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 3072,
|
13 |
+
"label2id": {
|
14 |
+
"entailment": 0,
|
15 |
+
"not_entailment": 1
|
16 |
+
},
|
17 |
+
"layer_norm_eps": 1e-07,
|
18 |
+
"max_position_embeddings": 512,
|
19 |
+
"max_relative_positions": -1,
|
20 |
+
"model_type": "deberta-v2",
|
21 |
+
"norm_rel_ebd": "layer_norm",
|
22 |
+
"num_attention_heads": 12,
|
23 |
+
"num_hidden_layers": 12,
|
24 |
+
"pad_token_id": 0,
|
25 |
+
"pooler_dropout": 0,
|
26 |
+
"pooler_hidden_act": "gelu",
|
27 |
+
"pooler_hidden_size": 768,
|
28 |
+
"pos_att_type": [
|
29 |
+
"p2c",
|
30 |
+
"c2p"
|
31 |
+
],
|
32 |
+
"position_biased_input": false,
|
33 |
+
"position_buckets": 256,
|
34 |
+
"relative_attention": true,
|
35 |
+
"share_att_key": true,
|
36 |
+
"torch_dtype": "float32",
|
37 |
+
"transformers_version": "4.36.0",
|
38 |
+
"type_vocab_size": 0,
|
39 |
+
"vocab_size": 251000
|
40 |
+
}
|