Fix configs and functions for distributed rank checks#111
Open
int11 wants to merge 2 commits intoIntellindust-AI-Lab:mainfrom
Open
Fix configs and functions for distributed rank checks#111int11 wants to merge 2 commits intoIntellindust-AI-Lab:mainfrom
int11 wants to merge 2 commits intoIntellindust-AI-Lab:mainfrom
Conversation
Zoxive
approved these changes
Aug 22, 2025
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Include the following four modifications.
configs/base/dfine_hgnetv2.yml: The weight file is currently being downloaded to an unusual location. Modified so that it downloads directly into the project folder for convenience.
configs/deim_dfine/deim_hgnetv2_n_coco.yml: The output_dir in this file is inconsistent with other models. Updated to maintain consistency.
configs/deim_dfine/dfine_hgnetv2_n_coco.yml: This file is the only one that directly specifies total_batch_size in train_dataloader and val_dataloader. Because of this, as noted in the README, modifying total_batch_size in configs/base/dataloader.yml does not properly affect training when using deim_hgnetv2_n. Fixed for consistency with the intended configuration behavior.
engine/backbone/hgnetv2.py: When training with a single GPU, the rank-checking function for loading HGNetV2 pretraining weights causes an error. This has been corrected.