GitLab

* fix drop_path_rate in classification

* add ms in detection 3x config

* fix installation procedure explanation of segmentation. This is mainly cuda related issues.

* let it runnable segmentation inference for image directory too.

* update segmentation readme for image directory processing.

* add new line

Co-authored-by: bingwen.ai <bingwen.ai@bytedance.com>

* update configs

* update readme

* upload mask2former_beit.py

* Add InternImage-H + Mask2Former

* Update README.md

* Update configs and readme

* Update README_CN.md

Support openlane

* revise readme

* revise readme

* revise readme

* updata readme

* update readme

* update readme

Co-authored-by: zhiqil <zhiqil@nvidia.com>

* minor update on deepspeed

* support config offloading

* support remove center point for DCNV3

* dcnv3: update test

* remove center for DCNv3_pytorch

* fix pytorch version

* fix unit test

* add backward compatibility

* bump dcn version to 1.1

support occupancy prediction

* support ema for main_deepspeed.py, fix torch.distribute.launch

* minor fix

* update readme

* Update README.md

* Update README_CN.md

* support ema for main_deepspeed.py, fix torch.distribute.launch

* minor fix

* update readme

* support ema for main_deepspeed.py, fix torch.distribute.launch

* minor fix

feat: add zero shot instance seg of sam prompted by prediction boxes of detector based on InternImage