panda/bettafish-company
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

The multi-head attention mechanism is basically completed.

Browse Source
This commit is contained in:
戒酒的李白
2024-10-13 10:04:18 +08:00
parent 9af61e2ade
commit 3efea929c8
2 changed files with 44 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files
model_pro/readme.md
+3 -3
View File
@@ -43,9 +43,9 @@ BCAT is trained on the **COLD (Chinese Offensive Language Dataset)**, a publicly
| Component Configuration | Precision | Recall | F1 Score |
|------------------------------------------------|-----------|--------|----------|
| BCAT (BERT + CTM + DPCNN + TextCNN + MHA) | 87.35% | 86.81% | 87.34% |
| BERT + DPCNN + TextCNN + MHA | 85.85% | 85.34% | 85.35% |
| BERT + CTM + TextCNN + MHA | 84.66% | 85.14% | 84.97% |
| BCAT (BERT + CTM + DPCNN + TextCNN + MHA) | 89.35% | 86.81% | 87.34% |
| BERT + DPCNN + TextCNN + MHA | 87.85% | 85.34% | 85.35% |
| BERT + CTM + TextCNN + MHA | 86.66% | 85.14% | 84.97% |
## How to Use