Image Classification on CIFAR-10 (Accuracy)
99.3AccuracySVFit
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SVFitBackbone=ViT-Large, Memory=1351.5 MB2026.05 | 99.3 | — | — | — | |
| SVFTBackbone=ViT-Large, Memory=1355.8 MB2026.05 | 99.3 | — | — | — | |
| CERSABackbone=ViT-Large, Memory=1232.3 MB2026.05 | 99.3 | — | — | — | |
| FTBackbone=ViT-Large, Memory=4629.8 MB2026.05 | 99.2 | — | — | — | |
| LoRABackbone=ViT-Large, Memory=1229.9 MB2026.05 | 99.1 | — | — | — | |
| DINOV2Domain=Vision, Embeddings=DINOV2, Size (K)=602025.07 | 99 | — | 0.1 | — | |
| V-MoEShots=10-shot, Protocol=Linear probing, Backbone=ViT-B/162025.11 | 98.94 | — | — | — | |
| AdapterTuneBackbone=ViT-B/16, Trainable Parameters (%)=0.9%2026.03 | 98.9 | — | — | — | |
| PiSSABackbone=ViT-Large, Memory=1229.9 MB2026.05 | 98.8 | — | — | — | |
| ERMoEShots=10-shot, Protocol=Linear probing, Backbone=ViT-B/162025.11 | 98.79 | — | — | — | |
| E2E-FT + WSEBackbone=CLIP ViT-B/162024.11 | 98.7 | — | — | — | |
| LP-FT + WSEBackbone=CLIP ViT-B/162024.11 | 98.7 | — | — | — | |
| E2E-FTBackbone=CLIP ViT-B/162024.11 | 98.6 | — | — | — | |
| E2E-FT + OSEBackbone=CLIP ViT-B/162024.11 | 98.6 | — | — | — | |
| E2E-FT + VRFBackbone=CLIP ViT-B/162024.11 | 98.6 | — | — | — | |
| LP-FT + OSEBackbone=CLIP ViT-B/162024.11 | 98.6 | — | — | — | |
| LP-FT + VRFBackbone=CLIP ViT-B/162024.11 | 98.6 | — | — | — | |
| Ge²mS-T LargeTime-step=42026.04 | 98.59 | — | — | — | |
| LP-FTBackbone=CLIP ViT-B/162024.11 | 98.5 | — | — | — | |
| HIVEFoundation Model=CLIP2026.03 | 98.49 | — | — | — | |
| HIVEFoundation Model=SigLIP2026.03 | 98.45 | — | — | — | |
| Ge²mS-T BaseTime-step=42026.04 | 98.45 | — | — | — | |
| BaseFoundation Model=SigLIP2026.03 | 98.43 | — | — | — | |
| SAFoundation Model=SigLIP2026.03 | 98.42 | — | — | — | |
| SAFoundation Model=CLIP2026.03 | 98.37 | — | — | — | |
| BaseFoundation Model=CLIP2026.03 | 98.32 | — | — | — | |
| Ge²mS-T SmallTime-step=42026.04 | 98.28 | — | — | — | |
| E2E-FT + WSEBackbone=CLIP ViT-B/322024.11 | 98.2 | — | — | — | |
| CtrlAc(2)Backbone=WRN-28-10, Training Setup=Modified, DA Pool=Control, Number of Augmentations (N)=22026.03 | 98.19 | — | — | — | |
| CtrlAc(1)Backbone=WRN-28-10, Training Setup=Modified, DA Pool=Control, Number of Augmentations (N)=12026.03 | 98.14 | — | — | — | |
| LP-FT + WSEBackbone=CLIP ViT-B/322024.11 | 98.1 | — | — | — | |
| LP-FT + OSEBackbone=CLIP ViT-B/322024.11 | 98.1 | — | — | — | |
| LP-FT + VRFBackbone=CLIP ViT-B/322024.11 | 98.1 | — | — | — | |
| CtrlAc(3)Backbone=WRN-28-10, Training Setup=Modified, DA Pool=Control, Number of Augmentations (N)=32026.03 | 98.1 | — | — | — | |
| NLEEPTop-k=k=32022.07 | 98.02 | — | — | — | |
| SFDATop-k=k=32022.07 | 98.02 | — | — | — | |
| SFDAcomTop-k=k=32022.07 | 98.02 | — | — | — | |
| MARVISDomain=Vision, Embeddings=DINOV2, Size (K)=602025.07 | 98 | — | 0.1 | — | |
| TAwBackbone=WRN-28-10, Training Setup=Modified, DA Pool=Wide2026.03 | 97.96 | — | — | — | |
| SFDAcomTop-k=k=22022.07 | 97.91 | — | — | — | |
| E2E-FTBackbone=CLIP ViT-B/322024.11 | 97.9 | — | — | — | |
| E2E-FT + OSEBackbone=CLIP ViT-B/322024.11 | 97.9 | — | — | — | |
| LP-FTBackbone=CLIP ViT-B/322024.11 | 97.9 | — | — | — | |
| LogMETop-k=k=32022.07 | 97.88 | — | — | — | |
| LogMETop-k=k=22022.07 | 97.86 | — | — | — | |
| NLEEPTop-k=k=22022.07 | 97.86 | — | — | — | |
| SFDATop-k=k=22022.07 | 97.86 | — | — | — | |
| E2E-FT + VRFBackbone=CLIP ViT-B/322024.11 | 97.8 | — | — | — | |
| LoRSPPre-trained Backbone=ViT-B/322026.06 | 97.77 | — | — | — | |
| LE-SAMBackbone=PyramidNet-1102026.05 | 97.63 | — | — | — | |
| DeepAAsBackbone=WRN-28-10, Training Setup=Standard, DA Pool=Standard2026.03 | 97.56 | — | — | — | |
| CtrlAc(2)Backbone=WRN-28-10, Training Setup=Standard, DA Pool=Control, Number of Augmentations (N)=22026.03 | 97.54 | — | — | — | |
| LoR-VPPre-trained Backbone=ViT-B/322026.06 | 97.52 | — | — | — | |
| AdapterTuneBackbone=ViT-S/16, Trainable Parameters (%)=0.9%2026.03 | 97.5 | — | — | — | |
| TAwBackbone=WRN-28-10, Training Setup=Standard, DA Pool=Wide2026.03 | 97.46 | — | — | — | |
| ASAMBackbone=PyramidNet-1102026.05 | 97.46 | — | — | — | |
| ESAMBackbone=PyramidNet-1102026.05 | 97.46 | — | — | — | |
| F-SAMBackbone=PyramidNet-1102026.05 | 97.46 | — | — | — | |
| LE-SAMBackbone=WRN-28-102026.05 | 97.44 | — | — | — | |
| Eigen-SAMBackbone=PyramidNet-1102026.05 | 97.44 | — | — | — | |
| AABackbone=WRN-28-10, Training Setup=Standard2026.03 | 97.4 | — | — | — | |
| Eigen-SAMBackbone=WRN-28-102026.05 | 97.36 | — | — | — | |
| F-SAMBackbone=WRN-28-102026.05 | 97.33 | — | — | — | |
| ASAMBackbone=WRN-28-102026.05 | 97.31 | — | — | — | |
| RAsBackbone=WRN-28-10, Training Setup=Standard, DA Pool=Standard2026.03 | 97.3 | — | — | — | |
| SAMBackbone=PyramidNet-1102026.05 | 97.3 | — | — | — | |
| ESAMBackbone=WRN-28-102026.05 | 97.29 | — | — | — | |
| SAMBackbone=WRN-28-102026.05 | 97.27 | — | — | — | |
| SeAl-KDArchitecture=ResNet-19, KD Setting=w/ KD, Timesteps (T)=62026.05 | 97.23 | — | — | — | |
| Full fine-tuningBackbone=ViT-S/16, Trainable Parameters (%)=100%2026.03 | 97.2 | — | — | — | |
| XSAMBackbone=DenseNet-121, Setting=Single-step2026.03 | 97.15 | — | — | — | |
| SeAl-KDArchitecture=ResNet-19, KD Setting=w/ KD, Timesteps (T)=42026.05 | 97.14 | — | — | — | |
| SpikformerType=Transfer Learning, Time-step=42026.04 | 97.03 | — | — | — | |
| ViT-B/16PEFT=-, Params(%)=86M2026.03 | 97 | — | — | — | |
| Logit-SNNArchitecture=ResNet-19, KD Setting=w/ KD, Timesteps (T)=62026.05 | 97 | — | — | — | |
| SAMBackbone=DenseNet-121, Setting=Single-step2026.03 | 96.97 | — | — | — | |
| Logit-SNNArchitecture=ResNet-19, KD Setting=w/ KD, Timesteps (T)=42026.05 | 96.97 | — | — | — | |
| Sw-Tiprecision=FP16, mode=Vanilla2026.06 | 96.94 | — | — | — | |
| XSAMBackbone=ResNet-18, k=22026.03 | 96.81 | — | — | — | |
| XSAMBackbone=ResNet-18, k=42026.03 | 96.81 | — | — | — | |
| MSAMBackbone=ResNet-18, k=42026.03 | 96.8 | — | — | — | |
| MSAMBackbone=ResNet-18, k=22026.03 | 96.78 | — | — | — | |
| HTA-KLArchitecture=ResNet-19, KD Setting=w/ KD, Timesteps (T)=42026.05 | 96.76 | — | — | — | |
| XSAMBackbone=ResNet-18, Setting=Single-step2026.03 | 96.74 | — | — | — | |
| XSAMBackbone=ResNet-18, k=12026.03 | 96.74 | — | — | — | |
| LSAMBackbone=ResNet-18, k=42026.03 | 96.72 | — | — | — | |
| RDCNet2026.04 | 96.71 | — | — | — | |
| Full fine-tuningBackbone=DeiT-T, Trainable Parameters (%)=100%2026.03 | 96.7 | — | — | — | |
| SSCNet2026.04 | 96.69 | — | — | — | |
| LSAMBackbone=ResNet-18, k=22026.03 | 96.66 | — | — | — | |
| LE-SAMBackbone=ResNet-182026.05 | 96.6 | — | — | — | |
| SAMBackbone=ResNet-18, Setting=Single-step2026.03 | 96.59 | — | — | — | |
| LSAMBackbone=ResNet-18, k=12026.03 | 96.59 | — | — | — | |
| MSAMBackbone=ResNet-18, k=12026.03 | 96.59 | — | — | — | |
| ESAMBackbone=ResNet-182026.05 | 96.56 | — | — | — | |
| SGDBackbone=PyramidNet-1102026.05 | 96.56 | — | — | — | |
| F-SAMBackbone=ResNet-182026.05 | 96.55 | — | — | — | |
| ASAMBackbone=ResNet-182026.05 | 96.53 | — | — | — | |
| SAMBackbone=ResNet-182026.05 | 96.52 | — | — | — | |
| DAMSNet2026.04 | 96.51 | — | — | — |