Hugging Face's logo

IDEA-CCNL
/
Erlangshen-Longformer-330M

Transformers
PyTorch
Chinese
longformer
Inference Endpoints
Model card Files Community
roygan commited on
Commit
a50c01a
•
1 Parent(s): 05fd5be

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +31 -30
config.json CHANGED
@@ -1,30 +1,31 @@
1
- {
2
- "architectures": [
3
- "LongformerForPreTraining"
4
- ],
5
- "attention_probs_dropout_prob": 0.1,
6
- "attention_window": 128,
7
- "use_sparse_attention":true,
8
- "directionality": "bidi",
9
- "gradient_checkpointing": false,
10
- "hidden_act": "gelu",
11
- "hidden_dropout_prob": 0.1,
12
- "hidden_size": 1024,
13
- "initializer_range": 0.02,
14
- "intermediate_size": 4096,
15
- "layer_norm_eps": 1e-12,
16
- "max_position_embeddings": 4096,
17
- "model_type": "longformer",
18
- "num_attention_heads": 16,
19
- "num_hidden_layers": 24,
20
- "pad_token_id": 0,
21
- "pooler_fc_size": 1024,
22
- "pooler_num_attention_heads": 16,
23
- "pooler_num_fc_layers": 3,
24
- "pooler_size_per_head": 128,
25
- "pooler_type": "first_token_transform",
26
- "type_vocab_size": 4,
27
- "vocab_size": 12000
28
- }
29
-
30
-
 
 
1
+ {
2
+ "architectures": [
3
+ "LongformerForPreTraining"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "attention_window": 128,
7
+ "use_sparse_attention":true,
8
+ "directionality": "bidi",
9
+ "gradient_checkpointing": false,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 4096,
15
+ "layer_norm_eps": 1e-12,
16
+ "max_position_embeddings": 4096,
17
+ "model_type": "longformer",
18
+ "num_attention_heads": 16,
19
+ "num_hidden_layers": 24,
20
+ "pad_token_id": 0,
21
+ "pooler_fc_size": 1024,
22
+ "pooler_num_attention_heads": 16,
23
+ "pooler_num_fc_layers": 3,
24
+ "pooler_size_per_head": 128,
25
+ "pooler_type": "first_token_transform",
26
+ "tokenizer_class": "BertTokenizer",
27
+ "type_vocab_size": 4,
28
+ "vocab_size": 12000
29
+ }
30
+
31
+