Hugging Face's logo

junnyu
/
roformer_v2_chinese_char_base

Fill-Mask Transformers PyTorch Chinese roformer roformer-v2 tf2.0
Model card Files Community
junnyu commited on
Commit
8dbb1e5
•
1 Parent(s): ae7735d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -6
README.md CHANGED
@@ -22,16 +22,15 @@ https://github.com/JunnYu/RoFormer_pytorch
22
  | | iflytek | tnews | afqmc | cmnli | ocnli | wsc | csl |
23
  | :-----: | :-----: | :---: | :---: | :---: | :---: | :---: | :---: |
24
  | BERT | 60.06 | 56.80 | 72.41 | 79.56 | 73.93 | 78.62 | 83.93 |
25
- | RoBERTa | 60.64 | 58.06 | 74.05 | **81.24** | 76.00 | 87.50 | 84.50 |
26
  | RoFormer | 60.91 | 57.54 | 73.52 | 80.92 | 76.07 | 86.84 | 84.63 |
27
  | RoFormer-pytorch(本仓库代码) | 60.60 | 57.51 | 74.44 | 80.79 | 75.67 | 86.84 | 84.77 |
28
  | RoFormerV2<sup>*</sup> | 60.87 | 56.54 | 72.75 | 80.34 | 75.36 | 80.92 | 84.67 |
29
- | RoFormerV2<sup>*</sup>-pytorch(本仓库代码) | **62.87** | **59.03** | **76.20** | 80.85 | **79.73** | **87.82** | **91.87** |
30
  | GAU-α | 61.41 | 57.76 | 74.17 | 81.82 | 75.86 | 79.93 | 85.67 |
31
  | GAU-α-pytorch（Adafactor） | 61.18 | 57.52 | 73.42 | 80.91 | 75.69 | 80.59 | 85.5 |
32
  | GAU-α-pytorch（AdamW wd0.01 warm_decay0.1） | 60.68 | 57.95 | 73.08 | 81.02 | 75.36 | 81.25 | 83.93 |
33
-
34
-
35
 
36
 
37
  ### CLUE-1.0-test榜单分类任务结果，base版本。
@@ -39,9 +38,10 @@ https://github.com/JunnYu/RoFormer_pytorch
39
  | | iflytek | tnews | afqmc | cmnli | ocnli | wsc | csl |
40
  | :-----: | :-----: | :---: | :---: | :---: | :---: | :---: | :---: |
41
  | RoFormer-pytorch(本仓库代码) | 59.54 | 57.34 | 74.46 | 80.23 | 73.67 | 80.69 | 84.57 |
42
- | RoFormerV2<sup>*</sup>-pytorch(本仓库代码) | 63.15 | 58.24 | 75.42 | 80.59 | 74.17 | 83.79 | 83.73 |
43
- | GAU-α-pytorch（Adafactor） | 61.38 | 57.08 | 74.05 | 80.37 | 73.53 | 74.83 | 85.6 |
44
  | GAU-α-pytorch（AdamW wd0.01 warm_decay0.1） | 60.54 | 57.67 | 72.44 | 80.32 | 72.97 | 76.55 | 84.13 |
 
45
 
46
  ## pytorch & tf2.0使用
47
  ```python
22
  | | iflytek | tnews | afqmc | cmnli | ocnli | wsc | csl |
23
  | :-----: | :-----: | :---: | :---: | :---: | :---: | :---: | :---: |
24
  | BERT | 60.06 | 56.80 | 72.41 | 79.56 | 73.93 | 78.62 | 83.93 |
25
+ | RoBERTa | 60.64 | 58.06 | 74.05 | 81.24 | 76.00 | 87.50 | 84.50 |
26
  | RoFormer | 60.91 | 57.54 | 73.52 | 80.92 | 76.07 | 86.84 | 84.63 |
27
  | RoFormer-pytorch(本仓库代码) | 60.60 | 57.51 | 74.44 | 80.79 | 75.67 | 86.84 | 84.77 |
28
  | RoFormerV2<sup>*</sup> | 60.87 | 56.54 | 72.75 | 80.34 | 75.36 | 80.92 | 84.67 |
29
+ | RoFormerV2<sup>*</sup>-pytorch(本仓库代码) | **62.87** | 59.03 | **76.20** | 80.85 | 79.73 | 87.82 | **91.87** |
30
  | GAU-α | 61.41 | 57.76 | 74.17 | 81.82 | 75.86 | 79.93 | 85.67 |
31
  | GAU-α-pytorch（Adafactor） | 61.18 | 57.52 | 73.42 | 80.91 | 75.69 | 80.59 | 85.5 |
32
  | GAU-α-pytorch（AdamW wd0.01 warm_decay0.1） | 60.68 | 57.95 | 73.08 | 81.02 | 75.36 | 81.25 | 83.93 |
33
+ | RoFormerV2<sup>*</sup>-large-pytorch(本仓库代码) | 61.75 | **59.21** | 76.14 | **82.35** | **81.73** | **91.45** | 91.5 |
 
34
 
35
 
36
  ### CLUE-1.0-test榜单分类任务结果，base版本。
38
  | | iflytek | tnews | afqmc | cmnli | ocnli | wsc | csl |
39
  | :-----: | :-----: | :---: | :---: | :---: | :---: | :---: | :---: |
40
  | RoFormer-pytorch(本仓库代码) | 59.54 | 57.34 | 74.46 | 80.23 | 73.67 | 80.69 | 84.57 |
41
+ | RoFormerV2<sup>*</sup>-pytorch(本仓库代码) | **63.15** | 58.24 | 75.42 | 80.59 | 74.17 | 83.79 | 83.73 |
42
+ | GAU-α-pytorch（Adafactor） | 61.38 | 57.08 | 74.05 | 80.37 | 73.53 | 74.83 | **85.6** |
43
  | GAU-α-pytorch（AdamW wd0.01 warm_decay0.1） | 60.54 | 57.67 | 72.44 | 80.32 | 72.97 | 76.55 | 84.13 |
44
+ | RoFormerV2-large<sup>*</sup>-pytorch(本仓库代码) | 61.85 | **59.13** | **76.38** | **80.97** | **76.23** | **85.86** | 84.33 |
45
 
46
  ## pytorch & tf2.0使用
47
  ```python