facebookresearch / facebookresearch/detectron2
Support Detectron2 on AMD's ROCM platform
Open
contributions welcome
enhancement
- Dominant language
- Python
- Stars
- 34.7k
- Forks
- 7.9k
- PR merge metrics
- No merged PRs in 30d
Description
## 🚀 Feature
The proposed feature is to support GPU acceleration on AMD's ROCM stack.
We (AMD) have already ported pytorch and torchvision to support our ROCM stack, and all those changes are currently in the pytorch and torchvision master branches, so detectron2 is the next natural step.
This has been already implemented actually, I also just made a pull request.
Contributor guide
Research direction
No files, tests, or entry points are named. Review the existing implementation or pull request mentioned in the issue, then compare it with the PyTorch and torchvision ROCM support; the work is done when Detectron2 runs with AMD ROCM GPU acceleration.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 20/100