About the code

[RuoyuChen10]

Hi, I would like to ask if your code is already runnable. The examples in the jupyter notebook from a year ago seem to have many bugs. Thanks for your nice work!

[JensRahnfeld]

Dear @chanwkimlab & @iancovert ,

thank you very much for the great work. It really helped me in my research project back then. As a token of gratitude, i would like to contribute to the repository through this PR. It mainly

• updates the benchmarking notebook which enabled me to reproduce your results on ImageNette and Pets
• cleans up some code to improve readability.

@RuoyuChen10 i hope this helps.

Best,
Jens

[chanwkimlab]

Thanks you for your great interest in our work! I did not encounter significant bugs while using the benchmarking notebook back then, but I believe there could be some minor unknown bugs as I haven't tested it rigorously across multiple environments. Thank you so much, Jens, for the pull request with cleaned-up code. I'll review this and update accordingly.

[RuoyuChen10]

Thanks you for your great interest in our work! I did not encounter significant bugs while using the benchmarking notebook back then, but I believe there could be some minor unknown bugs as I haven't tested it rigorously across multiple environments. Thank you so much, Jens, for the pull request with cleaned-up code. I'll review this and update accordingly.

Can you offer an example with CLIP? I hope that the comparison of our approach with yours, which will be performed on multimodal VIT, will also help to publicize your work.