Text Classification on Yelp Review (test)
69.63Macro F1Prompt-based FT (soft)
Evaluation Results
| Method | Links | |
|---|---|---|
| Prompt-based FT (soft)Number of labeled samples=full, Backbone=RoBERTa-Large2023.05 | 69.63 | |
| Prompt-based FT (hard)Number of labeled samples=full, Backbone=RoBERTa-Large2023.05 | 68.71 | |
| Prompt-based FT (hard) + PCPNumber of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 63.15 | |
| Prompt-based FT (soft) + PCPNumber of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 62.92 | |
| ADAMATCHNumber of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 58.95 | |
| DASHNumber of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 58.85 | |
| FIXMATCHNumber of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 58.51 | |
| FLEXMATCHNumber of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 58.34 | |
| Prompt-based FT (hard)Number of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 58.27 | |
| Prompt-based FT (soft)Number of labeled samples=200, Backbone=RoBERTa-Large2023.05 | 57.12 | |
| Prompt-based FT (hard) + PCPNumber of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 52.92 | |
| Prompt-based FT (soft) + PCPNumber of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 50.66 | |
| FLEXMATCHNumber of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 50.6 | |
| ADAMATCHNumber of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 50.42 | |
| DASHNumber of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 47.44 | |
| FIXMATCHNumber of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 47.26 | |
| Prompt-based FT (hard)Number of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 46.69 | |
| Prompt-based FT (soft)Number of labeled samples=40, Backbone=RoBERTa-Large2023.05 | 45.43 |