Romain Beaumont
Romain Beaumont
If you want help you need to describe exactly everything you did with all details, what you expected, and what didn't work when you tried to fix the issue. On...
Tested all models, it works for me
Tested all models, it works for me
Interesting! I'm experimenting with a 8 a100 pod. On one GPU I saw these numbers: * M2M small : 18 sample/s * Mbart : 20 sample/s * Opus : 40...
Quite possible. There is no solution to this problem yet though afaik
If you want them make them, there are no such pretrained model in existence with this repo
I think there is a confusion here :) This repo is a reproduction repo. The author of the repo does not have any pretrained model. If you want to reproduce...
if someone feels like plugging https://github.com/lucidrains/CoCa-pytorch/blob/main/coca_pytorch/coca_pytorch.py into https://github.com/mlfoundations/open_clip code base, I could quickly train it
LiT: Zero-Shot Transfer with Locked-image Text Tuning https://arxiv.org/abs/2111.07991 and in particular I think it would be interesting to be able to somehow transfer weights of existing models (clip image and...
https://github.com/facebookresearch/SLIP they combine the losses of CLIP (vision+language) and SimCLR (vision) and get better zero shot accuracy on a 15M dataset than clip on the same dataset Hopefully accuracies would...