search

SHI-Labs / Neighborhood-Attention-Transformer

Neighborhood Attention Transformer, arxiv 2022 / CVPR 2023. Dilated Neighborhood Attention Transformer, arxiv 2022
MIT License
1.04k stars 85 forks source link

How to train without the cuda extension? #10

Closed Geek-lixiang closed 2 years ago

Geek-lixiang commented 2 years ago

I can not use the cuda extension, how can I train the model without it? Thank you very much!

alihassanijr commented 2 years ago

Thank you for your interest. Unfortunately we don't have a cpu version yet. Were you unable to install the extension, or do you not have CUDA?

AntonyDrovalov commented 2 years ago

Hello! Can I run inference without using a cuda extension? (for example, if I have macos operating system)

alihassanijr commented 2 years ago

Unfortunately that is not possible at the moment because we don't have a cpu implementation. We plan to release one in the future.

AntonyDrovalov commented 2 years ago

Unfortunately that is not possible at the moment because we don't have a cpu implementation. We plan to release one in the future.

Thanks for reply. Looking forward to this update.

Geek-lixiang commented 2 years ago

Thank you very much for your reply.

alihassanijr commented 2 years ago

We released the pure PyTorch version (PR #23 ) along with a new version of the kernel. I'd like to thank @qwopqwop200 for their implementation of our RPB, which was used instead of our own in the PyTorch version.