Language Modeling on Tiny-Shakespeare (Accuracy, Communication, Computation)
57.93Accuracy (Test)Original
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| OriginalStage=After Unlearning2024.08 | 57.93 | 61.56 | 66.42 | 5,360,000,000 | 1,260,000,000,000,000 | |
| OriginalStage=After Recovery2024.08 | 57.93 | 61.56 | 66.42 | 5,360,000,000 | 1,260,000,000,000,000 | |
| FedQUITStage=After Recovery2024.08 | 57.9 | 61.53 | 57.9 | 456,000,000 | 107,000,000,000,000 | |
| FedAUStage=After Recovery2024.08 | 57.82 | 60.95 | 60.12 | 4,720,000,000 | 1,110,000,000,000,000 | |
| FedOSDStage=After Recovery2024.08 | 57.82 | 60.81 | 60.8 | 2,570,000,000 | 607,000,000,000,000 | |
| MoDeStage=After Recovery2024.08 | 57.75 | 60 | 61.3 | 2,790,000,000 | 657,000,000,000,000 | |
| NoTStage=After Recovery2024.08 | 57.51 | 58.31 | 59.9 | 6,430,000,000 | 1,520,000,000,000,000 | |
| RetrainStage=After Unlearning2024.08 | 57.43 | 61.29 | 58.6 | 5,360,000,000 | 1,260,000,000,000,000 | |
| RetrainStage=After Recovery2024.08 | 57.43 | 61.29 | 58.6 | 5,360,000,000 | 1,260,000,000,000,000 | |
| FedQUITStage=After Unlearning2024.08 | 53.18 | 56.29 | 49.1 | 456,000,000 | 107,000,000,000,000 | |
| FedOSDStage=After Unlearning2024.08 | 53.1 | 55.9 | 51.33 | 2,570,000,000 | 607,000,000,000,000 | |
| MoDeStage=After Unlearning2024.08 | 50.92 | 53.55 | 56.38 | 2,790,000,000 | 657,000,000,000,000 | |
| FedAUStage=After Unlearning2024.08 | 14.1 | 15.2 | 14.3 | 4,720,000,000 | 1,110,000,000,000,000 | |
| NoTStage=After Unlearning2024.08 | 1 | 1.2 | 1.3 | 6,430,000,000 | 1,520,000,000,000,000 |