Hezar: Upload training files
Browse files- model_config.yaml +36 -0
model_config.yaml
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: bert_sequence_labeling
|
| 2 |
+
config_type: model
|
| 3 |
+
task: sequence_labeling
|
| 4 |
+
num_labels: 13
|
| 5 |
+
id2label:
|
| 6 |
+
0: O
|
| 7 |
+
1: B-pro
|
| 8 |
+
2: I-pro
|
| 9 |
+
3: B-pers
|
| 10 |
+
4: I-pers
|
| 11 |
+
5: B-org
|
| 12 |
+
6: I-org
|
| 13 |
+
7: B-loc
|
| 14 |
+
8: I-loc
|
| 15 |
+
9: B-fac
|
| 16 |
+
10: I-fac
|
| 17 |
+
11: B-event
|
| 18 |
+
12: I-event
|
| 19 |
+
vocab_size: 42000
|
| 20 |
+
hidden_size: 768
|
| 21 |
+
num_hidden_layers: 12
|
| 22 |
+
num_attention_heads: 12
|
| 23 |
+
intermediate_size: 3072
|
| 24 |
+
hidden_act: gelu
|
| 25 |
+
hidden_dropout_prob: 0.1
|
| 26 |
+
attention_probs_dropout_prob: 0.1
|
| 27 |
+
max_position_embeddings: 512
|
| 28 |
+
type_vocab_size: 2
|
| 29 |
+
initializer_range: 0.02
|
| 30 |
+
layer_norm_eps: 1.0e-12
|
| 31 |
+
pad_token_id: 0
|
| 32 |
+
position_embedding_type: absolute
|
| 33 |
+
use_cache: true
|
| 34 |
+
prediction_skip_tokens:
|
| 35 |
+
- '[SEP]'
|
| 36 |
+
- '[CLS]'
|