First thanks a lot for releasing the code and the exhaustive instructions ! I've noticed quite a few people are trying to use meta-dataset under pytorch. It can however be cumbersome to try to integrate a TF based code into a pytorch one (even if doable). On my end, I decided to directly make a fully pytorch (minimalist) implementation of meta-dataset here, which has saved me quite some trouble so far ( you will find a full example + info on the structure of the code in the README.md). Hope it can help at least some people out there struggling with interfacing tensorflow and pytorch !
Hi,
First thanks a lot for releasing the code and the exhaustive instructions ! I've noticed quite a few people are trying to use meta-dataset under pytorch. It can however be cumbersome to try to integrate a TF based code into a pytorch one (even if doable). On my end, I decided to directly make a fully pytorch (minimalist) implementation of meta-dataset here, which has saved me quite some trouble so far ( you will find a full example + info on the structure of the code in the README.md). Hope it can help at least some people out there struggling with interfacing tensorflow and pytorch !
My best, Malik