Class-incremental learning on ImageNet100
92.47Average AccuracyJanus-LoRA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Janus-LoRAMode=Class-incremental2026.05 | 92.47 | 94.32 | 3.24 | |
| LoRA-DRSMode=Class-incremental2026.05 | 92.42 | 94.11 | 2.03 | |
| InfLoRAMode=Class-incremental2026.05 | 91.84 | 93.94 | 2.37 | |
| BiLoRAMode=Class-incremental2026.05 | 91.74 | 93.65 | 2.8 | |
| Coda-PromptMode=Class-incremental2026.05 | 91.63 | 93.93 | 3.12 | |
| LoRA-GPMMode=Class-incremental2026.05 | 91.53 | 94 | 3.58 | |
| L2PMode=Class-incremental2026.05 | 90.66 | 92.87 | 3.02 | |
| Fine-TuningMode=Class-incremental2026.05 | 89.1 | 92.95 | 6.8 | |
| Dual-PromptMode=Class-incremental2026.05 | 89.08 | 91.79 | 3.58 | |
| LingoCLBuffer size=0, K=32, B=50, C=10, Baseline=LUCIR2024.03 | 46.3 | 20.4 | 41.6 | |
| LingoCLBuffer size=0, K=16, B=50, C=10, Baseline=LUCIR2024.03 | 45.4 | 20 | 35.9 | |
| LingoCLBuffer size=1, K=8, B=50, C=10, Baseline=LUCIR2024.03 | 44.9 | 21.8 | 13.3 | |
| LingoCLBuffer size=1, K=32, B=50, C=10, Baseline=LUCIR2024.03 | 44.9 | 29.2 | 18.9 | |
| LingoCLBuffer size=1, K=4, B=50, C=10, Baseline=LUCIR2024.03 | 41.6 | 19.3 | 10.6 | |
| LUCIRBuffer size=1, K=8, B=50, C=102024.03 | 40.2 | 18.7 | 35 | |
| LUCIRBuffer size=1, K=4, B=50, C=102024.03 | 39.1 | 15.2 | 29.1 | |
| LingoCLBuffer size=1, K=16, B=50, C=10, Baseline=LUCIR2024.03 | 38.9 | 24.7 | 13.8 | |
| LUCIRBuffer size=1, K=32, B=50, C=102024.03 | 38.7 | 23.9 | 37.9 | |
| LingoCLBuffer size=0, K=8, B=50, C=10, Baseline=LUCIR2024.03 | 36.9 | 12.7 | 38.1 | |
| LUCIRBuffer size=0, K=32, B=50, C=102024.03 | 32.3 | 12.9 | 57.4 | |
| LUCIRBuffer size=1, K=16, B=50, C=102024.03 | 31.1 | 19.2 | 31.7 | |
| LUCIRBuffer size=0, K=16, B=50, C=102024.03 | 30.9 | 12.8 | 53.1 | |
| LUCIRBuffer size=0, K=8, B=50, C=102024.03 | 23.1 | 10.7 | 52.3 | |
| LUCIRBuffer size=0, K=4, B=50, C=102024.03 | 13.9 | 7.3 | 40.9 | |
| LingoCLBuffer size=0, K=4, B=50, C=10, Baseline=LUCIR2024.03 | 13.9 | 6 | 30.2 |