Natural Language Processing on Federated Dataset 1 Test-Time Personalization
78.1Paraphrase AccuracyFedDPA-F
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FedDPA-Fadapter type=local fine-tuning adapter2024.03 | 78.1 | 77.36 | 77.18 | 76.98 | 77.11 | 76.23 | 76.84 | 77.19 | 77.12 | |
| FedDPA-Tadapter type=local training adapter2024.03 | 76.2 | 75.51 | 76.19 | 75.63 | 74.86 | 74.6 | 74.77 | 75.96 | 75.47 | |
| FedLORA2024.03 | 75.56 | 76.55 | 75.21 | 74.94 | 76.16 | 74.64 | 74.99 | 76.97 | 75.63 | |
| Local-finetuned2024.03 | 48.99 | 47.24 | 27.53 | 22.66 | 48.86 | 49.07 | 46.45 | 52.09 | 42.86 |