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

Fix requirements #68

Closed alihassanijr closed 1 year ago

alihassanijr commented 1 year ago

This modifies README files in classification/, detection/ and segmentation/, and splits requirements.txt in each into two: requirements-base.txt and requirements.txt.

Reason: NATTEN requires PyTorch to be installed first, even when it's being set up with wheels, since it checks the torch version to ensure it matches up with the minimum requirements. pip install -r $FILE prepares packages and installs all of them if and only if all are set up successfully. This would fail to work for anyone that doesn't have torch preinstalled.

README files are modified to clarify this, upgrade to use the latest NATTEN, and to include the MMCV wheel URL.