Amin Rezaei

Testing properly is essential for such a productional smart contracts and all methods should be battle-tested. We can achieve this easily using TVM locally to ensure appropriate behavior of functions...

Thanks, I agree. However, increasing test coverage is essential to find bugs that are not simply visible to the eye. Also, the contracts didn't have major security issues. That was...

This is because weight off-loading is done per module rather than in batches. Full GPU use would significantly boost the speed. However, its implementation is not simple, and some calculations...

Thanks! I am delighted to hear that. Please do not hesitate to ask questions and feel free to contribute.

Hi, You're welcome Please provide more details on the weights you're using. The partial loader is aimed to load the large .pt or .bin file that can't be loaded directly...

If the model doesn't fit in your systems RAM and VRAM, then I'm afraid you wouldn't be able to use the split checkpoints with PyTorch-LIT as it's currently not supported....

Really nice implementation @Birch-san. Thanks for your detailed description. I'd appreciate it if you could make the modifications described above in this repo too and create a PR. Is it...

@brkirch Thanks for your interest and I'm glad this library is useful for you. This project is licensed under `MIT`, I updated the repository too.