Image Classification on CIFAR-100 10 samples/class
88.1Top-1 AccuracyLoRA (r=4)
Evaluation results
| Method | Links | |
|---|---|---|
| LoRA (r=4)Model Architecture=ViT-L, # Parameters=197K2026.04 | 88.1 | |
| LoRA (r=2)Model Architecture=ViT-L, # Parameters=98K2026.04 | 88 | |
| SOLAR r=4(4→1.6)Model Architecture=ViT-L, # Parameters=82K2026.04 | 87.9 | |
| NOLAModel Architecture=ViT-L, # Parameters=96K2026.04 | 87.8 | |
| SOLAR NOLA(4→1.2)Model Architecture=ViT-L, # Parameters=64K2026.04 | 87.7 | |
| SOLAR r=2(1→0.3)Model Architecture=ViT-L, # Parameters=50K2026.04 | 87.4 | |
| Full-FTModel Architecture=ViT-L, # Parameters=303M2026.04 | 86.2 | |
| LoRA (r=4)Model Architecture=ViT-B, # Parameters=74K2026.04 | 81.8 | |
| SOLAR r=4(4→1.6)Model Architecture=ViT-B, # Parameters=41K2026.04 | 81.5 | |
| NOLAModel Architecture=ViT-B, # Parameters=48K2026.04 | 81.3 | |
| SOLAR NOLA(4→1.2)Model Architecture=ViT-B, # Parameters=32K2026.04 | 81.1 | |
| Full-FTModel Architecture=ViT-B, # Parameters=86M2026.04 | 78.2 |