Skip to content

GitLab

Switch branch/tag
Find file Normal viewHistoryPermalink
contents.rst 3.44 KB
Newer Older
yuguo-Jack's avatar
llama_paddle  
yuguo-Jack committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29 


------------------------------------
SKEP模型汇总
------------------------------------



下表汇总介绍了目前PaddleNLP支持的SKEP模型对应预训练权重。
关于模型的具体细节可以参考对应链接。

+----------------------------------------------------------------------------------+--------------+----------------------------------------------------------------------------------+
| Pretrained Weight                                                                | Language     | Details of the model                                                             |
+==================================================================================+==============+==================================================================================+
|``skep_ernie_1.0_large_ch``                                                       | Chinese      | 24-layer, 1024-hidden,                                                           |
|                                                                                  |              | 16-heads, 336M parameters.                                                       |
|                                                                                  |              | Trained using the Erine model                                                    |
|                                                                                  |              | ``ernie_1.0``                                                                    |
+----------------------------------------------------------------------------------+--------------+----------------------------------------------------------------------------------+
|``skep_ernie_2.0_large_en``                                                       | English      | 24-layer, 1024-hidden,                                                           |
|                                                                                  |              | 16-heads, 336M parameters.                                                       |
|                                                                                  |              | Trained using the Erine model                                                    |
|                                                                                  |              | ``ernie_2.0_large_en``                                                           |
+----------------------------------------------------------------------------------+--------------+----------------------------------------------------------------------------------+
|``skep_roberta_large_en``                                                         | English      | 24-layer, 1024-hidden,                                                           |
|                                                                                  |              | 16-heads, 355M parameters.                                                       |
|                                                                                  |              | Trained using the RoBERTa model                                                  |
|                                                                                  |              | ``roberta_large_en``                                                             |
+----------------------------------------------------------------------------------+--------------+----------------------------------------------------------------------------------+