Textual Explanation on ImageNet-1K misclassified cases
1.7Mean ScoreFaithTrace
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FaithTraceModel=DINO ResNet-50, Setting=top-32026.05 | 1.7 | 0 | 2.971 | 8.14 | |
| FaithTraceModel=DINO ResNet-50, setting=top-52026.05 | 1.574 | 0.002 | 2.714 | 7.603 | |
| TEXTERModel=DINO ResNet-50, Setting=top-32026.05 | 0.668 | 0.22 | -0.002 | 3.44 | |
| TEXTERModel=DINO ResNet-50, setting=top-52026.05 | 0.643 | 0.232 | -0.141 | 3.464 | |
| Text-To-ConceptModel=DINO ResNet-50, setting=top-52026.05 | 0.507 | 0.248 | -0.281 | 3.189 | |
| Text-To-ConceptModel=DINO ResNet-50, Setting=top-32026.05 | 0.502 | 0.263 | -0.024 | 3.118 | |
| FaithTraceModel=ResNet-18, Setting=top-32026.05 | 0.19 | 0 | 2.496 | 4.869 | |
| FaithTraceModel=ResNet-18, setting=top-52026.05 | 0.171 | 0 | 2.271 | 4.598 | |
| FaithTraceModel=ViT, Setting=top-32026.05 | 0.13 | 0 | 0.874 | 1.077 | |
| FaithTraceModel=ViT, setting=top-52026.05 | 0.121 | 0 | 0.779 | 0.967 | |
| FaithTraceModel=ResNet-50, Setting=top-32026.05 | 0.069 | 0.007 | 1.375 | 1.732 | |
| TEXTERModel=ResNet-18, Setting=top-32026.05 | 0.068 | 0.277 | 0.579 | 2.571 | |
| TEXTERModel=ResNet-18, setting=top-52026.05 | 0.063 | 0.284 | 0.652 | 2.65 | |
| FaithTraceModel=ResNet-50, setting=top-52026.05 | 0.062 | 0.014 | 1.319 | 1.656 | |
| Text-To-ConceptModel=ResNet-18, Setting=top-32026.05 | 0.053 | 0.29 | -0.172 | 1.516 | |
| Text-To-ConceptModel=ResNet-18, setting=top-52026.05 | 0.051 | 0.314 | -0.184 | 1.552 | |
| FaithTraceModel=DINO ViT-S/8, Setting=top-32026.05 | 0.044 | 0.02 | 3.108 | 7.751 | |
| Text-To-ConceptModel=ViT, Setting=top-32026.05 | 0.043 | 0.267 | 0.184 | 0.303 | |
| TEXTERModel=ViT, Setting=top-32026.05 | 0.043 | 0.283 | 0.389 | 0.537 | |
| Text-To-ConceptModel=ViT, setting=top-52026.05 | 0.039 | 0.296 | 0.165 | 0.285 | |
| FaithTraceModel=DINO ViT-S/8, setting=top-52026.05 | 0.039 | 0.048 | 2.806 | 7.435 | |
| TEXTERModel=ViT, setting=top-52026.05 | 0.038 | 0.3 | 0.328 | 0.465 | |
| TEXTERModel=ResNet-50, Setting=top-32026.05 | 0.035 | 0.183 | 0.672 | 0.947 | |
| TEXTERModel=ResNet-50, setting=top-52026.05 | 0.03 | 0.23 | 0.598 | 0.853 | |
| Text-To-ConceptModel=ResNet-50, Setting=top-32026.05 | 0.021 | 0.253 | 0.253 | 0.477 | |
| Text-To-ConceptModel=ResNet-50, setting=top-52026.05 | 0.019 | 0.286 | 0.229 | 0.436 | |
| Text-To-ConceptModel=DINO ViT-S/8, Setting=top-32026.05 | 0.002 | 0.497 | -2.069 | 2.989 | |
| Text-To-ConceptModel=DINO ViT-S/8, setting=top-52026.05 | 0 | 0.524 | -2.127 | 2.77 | |
| TEXTERModel=DINO ViT-S/8, Setting=top-32026.05 | -0.01 | 0.623 | -1.961 | 2.405 | |
| TEXTERModel=DINO ViT-S/8, setting=top-52026.05 | -0.011 | 0.63 | -2.072 | 2.33 | |
| RandomModel=DINO ViT-S/8, Setting=top-32026.05 | -0.019 | 0.737 | -1.752 | 2.633 | |
| RandomModel=ViT, setting=top-52026.05 | -0.022 | 0.638 | 0.161 | 0.276 | |
| RandomModel=DINO ViT-S/8, setting=top-52026.05 | -0.024 | 0.74 | -2.156 | 2.237 | |
| RandomModel=ResNet-50, setting=top-52026.05 | -0.026 | 0.712 | 0.339 | 0.51 | |
| RandomModel=ResNet-50, Setting=top-32026.05 | -0.027 | 0.713 | 0.24 | 0.434 | |
| RandomModel=ViT, Setting=top-32026.05 | -0.028 | 0.657 | 0.187 | 0.293 | |
| RandomModel=ResNet-18, setting=top-52026.05 | -0.042 | 0.68 | 0.135 | 1.518 | |
| RandomModel=ResNet-18, Setting=top-32026.05 | -0.043 | 0.647 | 0.093 | 1.575 | |
| RandomModel=DINO ResNet-50, setting=top-52026.05 | -0.212 | 0.604 | -0.109 | 3.319 | |
| RandomModel=DINO ResNet-50, Setting=top-32026.05 | -0.243 | 0.593 | -0.001 | 3.283 |