Continual Learning on Evidence-sensitive stream
65.28Average ScoreRCL
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RCLBackbone=LLaVA-1.5-13B2026.07 | 65.28 | 3.76 | 0.093 | 13 | 11.1 | 90.6 | |
| RCLBackbone=InternVL-Chat-7B2026.07 | 64.75 | 3.93 | 0.098 | 13.7 | 11.8 | 89.9 | |
| RCLBackbone=LLaVA-1.5-7B2026.07 | 64.06 | 4.22 | 0.106 | 14.8 | 12.9 | 89 | |
| ProgLoRABackbone=LLaVA-1.5-7B2026.07 | 61.94 | 5.87 | 0.176 | 24.1 | 28.8 | 79.6 | |
| Ask-and-RememberBackbone=LLaVA-1.5-7B2026.07 | 61.9 | 5.8 | 0.159 | 21.6 | 24 | 82.9 | |
| GIFTBackbone=LLaVA-1.5-7B2026.07 | 61.62 | 6.06 | 0.162 | 21.9 | 24.7 | 82.6 | |
| MLLM-CLBackbone=LLaVA-1.5-7B2026.07 | 60.29 | 7.02 | 0.167 | 22.7 | 26.3 | 81.2 | |
| SEFEBackbone=LLaVA-1.5-7B2026.07 | 59.41 | 8.1 | 0.173 | 23.3 | 27.6 | 80.3 | |
| BranchLoRABackbone=LLaVA-1.5-7B2026.07 | 58.58 | 8.95 | 0.191 | 26.8 | 35 | 77.3 | |
| HiDe-LLaVABackbone=LLaVA-1.5-7B2026.07 | 57.26 | 9.38 | 0.196 | 28 | 36.3 | 76 | |
| ModalPromptBackbone=LLaVA-1.5-7B2026.07 | 56.99 | 9.12 | 0.184 | 25 | 31.4 | 78.2 | |
| Continual LLaVABackbone=LLaVA-1.5-7B2026.07 | 56.53 | 9.86 | 0.209 | 29.3 | 38.6 | 74.7 | |
| O-LoRABackbone=LLaVA-1.5-7B2026.07 | 54.3 | 12.6 | 0.234 | 33.9 | 46.7 | 70.7 | |
| MoELoRABackbone=LLaVA-1.5-7B2026.07 | 53.62 | 13.22 | 0.223 | 32.6 | 45.4 | 71.3 | |
| EWCBackbone=LLaVA-1.5-7B2026.07 | 51.96 | 16.06 | 0.257 | 36.7 | 54 | 67.4 | |
| LwFBackbone=LLaVA-1.5-7B2026.07 | 50.88 | 16.45 | 0.25 | 36.1 | 52.9 | 67.1 |