Image Classification on CIFAR100 (Accuracy, Uncertainty)
88.92AccuracyMMFT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MMFTModel=ViT, Method=Ours2026.01 | 88.92 | — | |
| FLYPModel=ViT, Method=FLYP2026.01 | 88.56 | — | |
| FTModel=ViT, Method=FT2026.01 | 88.23 | — | |
| StandardBackbone=ViT-L/142025.11 | 86.99 | — | |
| FCBMBackbone=ViT-L/142025.11 | 83.63 | — | |
| MMFTModel=RN50, Method=Ours2026.01 | 82.55 | — | |
| CF-CBMBackbone=ViT-L/142025.11 | 82.33 | — | |
| FLYPModel=RN50, Method=FLYP2026.01 | 82.18 | — | |
| LF-CBMBackbone=ViT-L/142025.11 | 81.98 | — | |
| FTModel=RN50, Method=FT2026.01 | 81.83 | — | |
| Standard (sparse)Backbone=ViT-L/142025.11 | 80.28 | — | |
| PC-DARTS-CDFLOPS(M)=5572021.07 | 80.01 | 0.964 | |
| µDARTSFLOPS(M)=6022021.07 | 79.98 | 0.561 | |
| ASAP-CDFLOPS(M)=5732021.07 | 79.27 | 0.992 | |
| SCS-SupConBackbone=ResNet502025.12 | 79.2 | — | |
| GOLD-NAS-I-CDFLOPS(M)=4632021.07 | 79.15 | 1.003 | |
| FNCLBackbone=ResNet502025.12 | 79.1 | — | |
| GOLD-NAS-A-CDFLOPS(M)=2782021.07 | 78.97 | 0.992 | |
| P-DARTS-CDFLOPS(M)=5322021.07 | 78.94 | 0.883 | |
| CSTCNBackbone=ResNet502025.12 | 78.7 | — | |
| TimeSCLBackbone=ResNet502025.12 | 78.7 | — | |
| DACLBackbone=ResNet502025.12 | 78.7 | — | |
| CSA-RSICBackbone=ResNet502025.12 | 78.6 | — | |
| PCLBackbone=ResNet502025.12 | 78.5 | — | |
| Circle LossBackbone=ResNet502025.12 | 78.5 | — | |
| CS-SupCon w. ov.Backbone=ResNet502025.12 | 78.4 | — | |
| PaCoBackbone=ResNet502025.12 | 78.3 | — | |
| SelfConBackbone=ResNet502025.12 | 78.1 | — | |
| CS-SupConBackbone=ResNet502025.12 | 77.6 | — | |
| RDARTS-CDFLOPS(M)=6052021.07 | 77.34 | 0.815 | |
| EfficientNet-B0FLOPS(M)=3902021.07 | 76.02 | 1.215 | |
| SupConBackbone=ResNet502025.12 | 75.5 | — | |
| DARTSFLOPS(M)=5952021.07 | 75.37 | — | |
| ResNet20FLOPS(M)=3202021.07 | 75.22 | 1.246 | |
| BYOLBackbone=ResNet502025.12 | 75 | — | |
| DARTS-CDFLOPS(M)=5982021.07 | 74.81 | 1.097 | |
| BaselineBackbone=ResNet502025.12 | 74.8 | — | |
| PCBMBackbone=ViT-L/142025.11 | 74.29 | — | |
| SimCLRBackbone=ResNet502025.12 | 73.9 | — | |
| VGG16FLOPS(M)=1602021.07 | 71.19 | 1.482 | |
| StandardBackbone=ResNet502025.11 | 70.19 | — | |
| ZSModel=MLLM, Method=ZS2026.01 | 70.09 | — | |
| MobileNetv2FLOPS(M)=5692021.07 | 68.54 | 1.266 | |
| FCBMBackbone=ResNet502025.11 | 64.77 | — | |
| LF-CBMBackbone=ResNet502025.11 | 64.62 | — | |
| CF-CBMBackbone=ResNet502025.11 | 64.31 | — | |
| ZSModel=ViT, Method=ZS2026.01 | 62.67 | — | |
| Standard (sparse)Backbone=ResNet502025.11 | 57.54 | — | |
| PCBMBackbone=ResNet502025.11 | 56.24 | — | |
| ZSModel=RN50, Method=ZS2026.01 | 40.98 | — |