fushh

Yes, according to this line, we can dynamically adjust the number of groups for each batch but not for each image. I think we can further utilize the memory by...

You can run with the following command: `python -m torch.distributed.launch --nproc_per_node=8 --use_env main.py --coco_path /YOUR_PATH/coco --resume ../r50_deformable_detr_plus_iterative_bbox_refinement_plus_plus_two_stage-checkpoint.pth --eval --with_box_refine --two_stage --num_queries 500` No other modification is needed.