Xingchao Peng

I am able to get parity result on pets dataset. But my cifar10 result is worse than yours. Refer to another issue I created: https://github.com/openai/CLIP/issues/164 | Model name | ResNet50...

> > I use the code in the `README.md (Zero-Shot Prediction)` to test the acc of ViT/B-32 on the CIFAR100 dataset and get the result of about 62% by top1...

The issue is the index of 'dinningtable' in prompts by the way

In addition, I also checked the implementation in [https://github.com/openai/CLIP/blob/fcab8b6eb92af684e7ff0a904464be7b99b49b88/notebooks/Prompt_Engineering_for_ImageNet.ipynb](https://github.com/openai/CLIP/blob/fcab8b6eb92af684e7ff0a904464be7b99b49b88/notebooks/Prompt_Engineering_for_ImageNet.ipynb) I changed the dataset from ImageNet in the notebook to CIFAR10, and got the result: 71.48%, which is very close to...