Continual Learning Classification on Standard CL Benchmark (4 tasks)
88.25Average Accuracy (AA)SAE-FD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SAE-FDBackbone=LLaMA-2-7B-Chat, Orders=32026.05 | 88.25 | 0.89 | |
| SeqLoRABackbone=LLaMA-2-7B-Chat, Orders=32026.05 | 85.62 | 4.71 | |
| SAPT-LoRABackbone=LLaMA-2-7B-Chat2026.05 | 81.1 | — | |
| SLAOBackbone=LLaMA-2-7B-Chat2026.05 | 80.4 | — | |
| DATABackbone=LLaMA-2-7B-Chat2026.05 | 79.8 | — | |
| InfLoRABackbone=LLaMA-2-7B-Chat2026.05 | 79.6 | — | |
| GORPBackbone=LLaMA-2-7B-Chat2026.05 | 78.6 | — | |
| N-LoRABackbone=LLaMA-2-7B-Chat2026.05 | 77.6 | — | |
| O-LoRABackbone=LLaMA-2-7B-Chat2026.05 | 76.1 | — |