Natural Language Processing on Federated Dataset 1 (Personalization)
0.805Paraphrasing ScoreFedDPA-T
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FedDPA-Tadapter type=local training adapter2024.03 | 0.805 | 0.845 | 0.7279 | 0.9651 | 0.735 | 0.62 | 0.7793 | 0.94 | 0.8022 | |
| FedDPA-Fadapter type=local fine-tuning adapter2024.03 | 0.79 | 0.845 | 0.7206 | 0.969 | 0.72 | 0.65 | 0.7386 | 0.925 | 0.7948 | |
| FedLORA2024.03 | 0.775 | 0.84 | 0.7149 | 0.9669 | 0.735 | 0.65 | 0.7527 | 0.92 | 0.7943 | |
| Centralized2024.03 | 0.77 | 0.82 | 0.7258 | 0.9659 | 0.705 | 0.635 | 0.7759 | 0.89 | 0.786 | |
| Local-finetuned2024.03 | 0.745 | 0.8 | 0.7371 | 0.9736 | 0.75 | 0.545 | 0.6855 | 0.895 | 0.7664 | |
| FedIT2024.03 | 0.69 | 0.83 | 0.7125 | 0.9632 | 0.715 | 0.625 | 0.7543 | 0.915 | 0.775 |