-
Notifications
You must be signed in to change notification settings - Fork 4.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Explicitly use the linalg.vector_norm call in comm/
#6960
Conversation
…orm` (#6931) - [x] Update PR since `torch.norm` and `torch.linalg.norm` have [different function signatures](https://pytorch.org/docs/stable/generated/torch.linalg.norm.html#torch.linalg.norm). - [x] Check if there are any numeric differences between the functions. - [x] Determine why there appear to be performance differences from others [here](pytorch/pytorch#136360). - [x] Update to `torch.linalg.vectornorm` Follow up PR handles these in the comm folder: #6960
Hi @loadams |
@GuanhuaWang - this is from the pytorch docs, it was mostly to make it more clear, but we don't "need" to update these like we needed to update from ![]() |
…orm` (deepspeedai#6931) - [x] Update PR since `torch.norm` and `torch.linalg.norm` have [different function signatures](https://pytorch.org/docs/stable/generated/torch.linalg.norm.html#torch.linalg.norm). - [x] Check if there are any numeric differences between the functions. - [x] Determine why there appear to be performance differences from others [here](pytorch/pytorch#136360). - [x] Update to `torch.linalg.vectornorm` Follow up PR handles these in the comm folder: deepspeedai#6960 Signed-off-by: siqi <[email protected]>
ds-chat workflow: https://github.com/microsoft/DeepSpeed/actions/runs/12835483030