Reproducing the results in the paper

lucidrains / CoCa-pytorch

Implementation of CoCa, Contrastive Captioners are Image-Text Foundation Models, in Pytorch

MIT License

1.04k stars 88 forks source link

Reproducing the results in the paper #5

Closed GKIBMNY closed 10 months ago

GKIBMNY commented 2 years ago

Thanks for this repo. Curious, is this an independent implementation of the CoCa paper? If yes, did you reproduce any result in the paper to ensure correctness of implementation?

Camellia-tx commented 10 months ago

Hi, I am also reproducing this model, did you succeed in reproducing it? I'm experiencing the problem that the contrastive loss never changes while training, have you encountered this problem?

lucidrains commented 10 months ago

@Camellia-tx they have already reproduced it over at OpenClip https://laion.ai/blog/coca/