Continual Learning on L1-L6 Overall Average (test)
79.2Average AccuracyGRID
Evaluation Results
| Method | Links | |
|---|---|---|
| GRIDBackbone=T5-large, Data Replay=false2025.07 | 79.2 | |
| SHLPTBackbone=T5-large, Data Replay=false2025.07 | 77.8 | |
| ProgPromptBackbone=T5-large, Data Replay=false2025.07 | 76 | |
| Per-task PromptBackbone=T5-large, Data Replay=false2025.07 | 75.2 | |
| LFPT5Backbone=T5-large, Data Replay=true2025.07 | 69.3 | |
| Data ReplayBackbone=T5-large, Data Replay=true2025.07 | 54.7 | |
| Prompt TuningBackbone=T5-large, Data Replay=false2025.07 | 9.1 | |
| FinetuneBackbone=T5-large, Data Replay=false2025.07 | 8.3 |