facebookresearch / facebookresearch/sam2
How to specify a set of GPUs not starting at index 0?
Open
- Dominant language
- Jupyter Notebook
- Stars
- 19.9k
- Forks
- 2.5k
- PR merge metrics
- No merged PRs in 30d
Description
All,
I have a number of GPUs in my server, however, it's a shared resource. I need to be able to specify which GPUs to use for my training. I have not found a straightforward way to configure the training to use "cuda:4" or the like.
What I need even more is to be able to use "cuda:4" through "cuda:9". I'm only running locally (i.e. no distributed nodes).
Is there a way to configure the above?
Contributor guide
Assessment
This issue has not been assessed yet.