search

postech-ami / Sound2Scene

26 stars 6 forks source link

Request for lnception-V3 finetuned checkpoint for image generation metrics and CLIP R@1,5 used checkpoint #2

Closed DragonLiu1995 closed 11 months ago

DragonLiu1995 commented 12 months ago

Hi,

Thanks for releasing the code for this amazing work! After I carefully read through the paper and supplementary materials, I have 2 questions.

  1. I found that you fine-tuned inception-V3 on VggSound data in order to calculate the FID and IS score. Could you please release the fine-tuned checkpoint so that I could use it to compare?

  2. Which specific checkpoint did you use for CLIP to calculate those CLIP R@1, 5 metrics?

Thanks in advance for guidance and clarifications!

sbkim052 commented 11 months ago

I updated the README for more information. Please check them:)