Image Classification on ImageNet (Accuracy Reported)
94.75AccuracyMoELoRA
Evaluation Results
| Method | Links | |
|---|---|---|
| MoELoRAevaluation_phase=Best during tuning2025.12 | 94.75 | |
| Upper bound2025.12 | 91.88 | |
| CL-MoEevaluation_phase=Best during tuning2025.12 | 90.08 | |
| Seq LoRAevaluation_phase=Best during tuning2025.12 | 89.41 | |
| DOT-CBM2025.05 | 83.84 | |
| LaBo2025.05 | 82.93 | |
| SparseCBM2025.05 | 82.85 | |
| CoopCBM2025.05 | 82.73 | |
| CEM2025.05 | 81.29 | |
| MoDEevaluation_phase=Best during tuning2025.12 | 80.67 | |
| Vanilla-CBM2025.05 | 79.17 | |
| Model tailorevaluation_phase=Best during tuning2025.12 | 78.22 | |
| DINOv2-ARLN=20M, Zero-shot=true2024.09 | 76.3 | |
| LiT L16LN=112M, Zero-shot=true2024.09 | 75.7 | |
| OpenAI-CLIP VIT-LN=400M, Zero-shot=true2024.09 | 75.3 | |
| DINOv2-MpNetN=20M, Zero-shot=true2024.09 | 74.8 | |
| MMRL + AmPLevariant=standard2025.12 | 73.17 | |
| MMAvariant=†2025.12 | 73.03 | |
| LAION-CLIP VIT-LN=400M, Zero-shot=true2024.09 | 72.7 | |
| MMA + AmPLevariant=standard2025.12 | 72.7 | |
| AttriPromptvariant=standard2025.12 | 72.4 | |
| IPOvariant=standard2025.12 | 72.15 | |
| MMRLmode=Source Evaluation2025.05 | 72.03 | |
| MMRLvariant=standard2025.12 | 72.03 | |
| MMRLvariant=†2025.12 | 72 | |
| MMRL++mode=Source Evaluation2025.05 | 71.87 | |
| CoOpmode=Source Evaluation2025.05 | 71.51 | |
| CoOpvariant=standard2025.12 | 71.51 | |
| TCPmode=Source Evaluation2025.05 | 71.4 | |
| PromptSRCmode=Source Evaluation2025.05 | 71.27 | |
| CoOpOpmode=Source Evaluation2025.05 | 71.02 | |
| CoCoOpvariant=standard2025.12 | 71.02 | |
| MMAmode=Source Evaluation2025.05 | 71 | |
| MMAvariant=standard2025.12 | 71 | |
| MaPLe + AmPLevariant=standard2025.12 | 70.87 | |
| MaPLemode=Source Evaluation2025.05 | 70.72 | |
| MaPLevariant=standard2025.12 | 70.72 | |
| MaPLevariant=†2025.12 | 70.5 | |
| CLIPvariant=standard2025.12 | 66.72 | |
| DualPromptevaluation_phase=Best during tuning2025.12 | 24.55 | |
| MoELoRAevaluation_phase=Final accuracy2025.12 | 18.73 | |
| Zero-shot2025.12 | 16.53 | |
| CL-MoEevaluation_phase=Final accuracy2025.12 | 15.62 | |
| Model tailorevaluation_phase=Final accuracy2025.12 | 14.97 | |
| MoDEevaluation_phase=Final accuracy2025.12 | 14.67 | |
| Seq LoRAevaluation_phase=Final accuracy2025.12 | 13.68 | |
| DualPromptevaluation_phase=Final accuracy2025.12 | 8.61 |