Are models on Pytorch Hub screened for errors?

I’m planning on using pytorch hub segmentation models for my research project. The models look fine overall, but I’m wondering if they have been screened for possible errors? I know that they are pooled in by other people on github, but is the pytorch team taking the time to screen them? Or is it an automated process with no checks?

Essentially, Do models on pytorch hub have high credibility, like tensorflow hub?

Thanks a lot for the initiative!

@ailzhang might know more about the process in PyTorch Hub.

Yes, absolutely.
We carefully vet and review them to the best extent we can.
If they are on they are screened.


Amazing! Thanks @smth @ptrblck, the research community thanks you for making our job easier through Pytorch in so many ways and continually improving it :slight_smile: