runtimeerror: tensors must be cuda and dense error with weighted image selection for training

[lleye]

I get a cuda runtime error when doing weighted image selection in ddp mode. below is how to replicate:

python -m torch.distributed.launch --nproc_per_node .. --image-weights leads to runtimeerror: tensors must be cuda and dense error.

Any idea why this is so?

[github-actions]

👋 Hello @lleye, thank you for your interest in 🚀 YOLOv5! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://www.ultralytics.com or email Glenn Jocher at glenn.jocher@ultralytics.com.

Requirements

Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), testing (test.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[lleye]

@glenn-jocher any thoughts

[glenn-jocher]

@lleye I think these two options may simply be incompatible with each other. DDP loses dataloader modification capabilities which --img-weights relies on.

If you have time you a PR with an assert to train to check for a combination of these two with an informative error message might help future users.

[glenn-jocher]

@lleye good news 😃! Your original issue may now been fixed ✅ in PR #3275. This provides improved error handling to notify the user than DDP is not compatible with the --image-weights training argument. To receive this update you can:

• git pull from within your yolov5/ directory
• git clone https://github.com/ultralytics/yolov5 again
• Force-reload PyTorch Hub: model = torch.hub.load('ultralytics/yolov5', 'yolov5s', force_reload=True)
• View our updated notebooks:

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 🚀!

[lleye]

thanks @glenn-jocher. any plans on making this compatible. if not, do you any suggestions on easily implementable methods for extreme class imbalance in yolov5?

[glenn-jocher]

@lleye reduce your imbalance with more data from underrepresented classes.

Also there's no requirement to train DDP, --image-weights works correctly in all other use cases.