Image Classification on UCF-101 (test)
90.67AccuracyLinear Probe
Evaluation Results
| Method | Links | |
|---|---|---|
| Linear ProbeInterpretability=false, Evaluation Setting=Fully-supervised, Backbone=CLIP ViT-L/142026.01 | 90.67 | |
| PCBM-ReDInterpretability=true, Evaluation Setting=Fully-supervised, Backbone=CLIP ViT-L/142026.01 | 90.38 | |
| LaBoInterpretability=true, Evaluation Setting=Fully-supervised, Backbone=CLIP ViT-L/142026.01 | 90.11 | |
| 2SFSBackbone=ViT-L/14, Shots=162025.03 | 89.5 | |
| Res-CBMInterpretability=true, Evaluation Setting=Fully-supervised, Backbone=CLIP ViT-L/142026.01 | 88.37 | |
| 2SFSBackbone=ViT-B/16, Shots=162025.03 | 86.6 | |
| 2SFSBackbone=ViT-B/32, Shots=162025.03 | 82.3 | |
| APArchitecture=CLIP (ViT-B/32)2026.04 | 76.42 | |
| CuPLInterpretability=false, Evaluation Setting=Zero-shot, Backbone=CLIP ViT-L/142026.01 | 76 | |
| APArchitecture=Swin-Transformer (Swin-B)2026.04 | 75.92 | |
| PCBM-ReD + CuPLInterpretability=true, Evaluation Setting=Zero-shot, Backbone=CLIP ViT-L/142026.01 | 75.91 | |
| PCBM-ReDInterpretability=true, Evaluation Setting=Zero-shot, Backbone=CLIP ViT-L/142026.01 | 75.36 | |
| CLIPInterpretability=false, Evaluation Setting=Zero-shot, Backbone=CLIP ViT-L/142026.01 | 75.31 | |
| VPArchitecture=Swin-Transformer (Swin-B)2026.04 | 74.41 | |
| HopSgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 74.2 | |
| HopSgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 73.8 | |
| HopSgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 73.1 | |
| HopSrandom confusion rate (gamma_c)=0.75, prompting=uni-prompt2026.04 | 72.9 | |
| DMN-ZSBackbone=ViT-B/162026.06 | 72.51 | |
| HopSgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 72.3 | |
| HopSrandom confusion rate (gamma_c)=0.50, prompting=uni-prompt2026.04 | 71.9 | |
| ComMemBackbone=ViT-B/162026.06 | 71.74 | |
| HopSrandom confusion rate (gamma_c)=0.80, prompting=uni-prompt2026.04 | 71.7 | |
| CCgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 71.4 | |
| SCEgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 71.3 | |
| HopSrandom confusion rate (gamma_c)=0.88, prompting=uni-prompt2026.04 | 71.2 | |
| TPSBackbone=ViT-B/162026.06 | 71 | |
| TDABackbone=ViT-B/162026.06 | 70.66 | |
| RCgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 70.5 | |
| DPEBackbone=ViT-B/162026.06 | 70.44 | |
| VPArchitecture=CLIP (ViT-B/32)2026.04 | 70.42 | |
| HopSgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 70.4 | |
| LWCgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 70.3 | |
| RCgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 70.3 | |
| MSEgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 69.9 | |
| HopSrandom confusion rate (gamma_c)=0.90, prompting=uni-prompt2026.04 | 69.2 | |
| SCEgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 69 | |
| CCgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 68.8 | |
| DynaPromptBackbone=ViT-B/162026.06 | 68.72 | |
| SCEgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 68.5 | |
| TPTBackbone=ViT-B/162026.06 | 68.04 | |
| SCEgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 68 | |
| SCEgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 66.8 | |
| CoOpBackbone=ViT-B/162026.06 | 66.55 | |
| LWCgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 65.6 | |
| DMN-ZSBackbone=ResNet-502026.06 | 65.34 | |
| RCgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 65.3 | |
| LWCgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 65.2 | |
| EnsembleBackbone=ViT-B/162026.06 | 65.16 | |
| CLIPBackbone=ViT-B/162026.06 | 65.13 | |
| TDABackbone=ResNet-502026.06 | 64.18 | |
| CCgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 63.3 | |
| DiffTPTBackbone=ResNet-502026.06 | 62.67 | |
| DiffTPTBackbone=ViT-B/162026.06 | 62.67 | |
| ComMemBackbone=ResNet-502026.06 | 62.62 | |
| DPEBackbone=ResNet-502026.06 | 61.98 | |
| CLIP-RefineBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 61.85 | |
| LWCgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 61.8 | |
| HyCDBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 61.7 | |
| HopSBackbone (Insd)=ResNet-18, Insd confusion rate (γc)=0.67, Prompting (uni-prompt)=true2026.04 | 60.9 | |
| TPTBackbone=ResNet-502026.06 | 60.82 | |
| MSEgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 60.7 | |
| Self-KDBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 60.16 | |
| Pre-trainedBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 59.9 | |
| DynaPromptBackbone=ResNet-502026.06 | 59.87 | |
| HopSBackbone (Insd)=ResNet-50, Insd confusion rate (γc)=0.67, Prompting (uni-prompt)=true2026.04 | 59.7 | |
| CCgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 59.6 | |
| EnsembleBackbone=ResNet-502026.06 | 59.48 | |
| MSEgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 59.3 | |
| RCgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 59.3 | |
| LWCgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 59.2 | |
| CoOpBackbone=ResNet-502026.06 | 59.05 | |
| CCgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 59 | |
| CLIPBackbone=ResNet-502026.06 | 58.84 | |
| m²-mixBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 58.76 | |
| HyCD + LalignBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 58.55 | |
| MSEgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 58.3 | |
| ContrastiveBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2025.04 | 58.26 | |
| MSEgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 54.3 | |
| HopSBackbone (Insd)=ResNet-18, Insd confusion rate (γc)=0.75, Prompting (uni-prompt)=true2026.04 | 53.5 | |
| HopSBackbone=ResNet-18, Confusion rate=0.80, Prompt type=uni-prompt2026.04 | 53.2 | |
| HopSBackbone (Insd)=CLIP-ResNet50, Insd confusion rate (γc)=0.67, Prompting (uni-prompt)=true2026.04 | 49.3 | |
| HopSBackbone (Insd)=ResNet-50, Insd confusion rate (γc)=0.75, Prompting (uni-prompt)=true2026.04 | 49.1 | |
| RCgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 49 | |
| HopSPrompting strategy=uni-prompt, Missing rate=25.0%, Label confusion setting=insd, Number of shots=16-shot2026.04 | 46 | |
| SCEPrompting strategy=uni-prompt, Missing rate=25.0%, Label confusion setting=insd, Number of shots=16-shot2026.04 | 45 | |
| HopSBackbone=ResNet-50, Confusion rate=0.80, Prompt type=uni-prompt2026.04 | 43 | |
| EXPgamma_c=0.75, prompt-mode=cls-prompts2026.04 | 42.3 | |
| CCPrompting strategy=uni-prompt, Missing rate=25.0%, Label confusion setting=insd, Number of shots=16-shot2026.04 | 41 | |
| EXPgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 39.3 | |
| EXPgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 39.3 | |
| HopSBackbone (Insd)=CLIP-ResNet50, Insd confusion rate (γc)=0.75, Prompting (uni-prompt)=true2026.04 | 38.4 | |
| EXPgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 37.9 | |
| MAEgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 37.7 | |
| LWCBackbone=ResNet-18, Confusion rate=0.80, Prompt type=uni-prompt2026.04 | 37.7 | |
| GCEgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 37.2 | |
| MAEgamma_c=0.80, prompt-mode=cls-prompts2026.04 | 34.9 | |
| EXPgamma_c=0.90, prompt-mode=cls-prompts2026.04 | 34.7 | |
| GCEgamma_c=0.88, prompt-mode=cls-prompts2026.04 | 34.6 | |
| GCEgamma_c=0.50, prompt-mode=cls-prompts2026.04 | 34.3 |