-
Notifications
You must be signed in to change notification settings - Fork 2.7k
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
Fix nemo 1 packed sequence TE version error #11874
Conversation
Signed-off-by: Chen Cui <[email protected]>
Is this a safe hacking? wouldn't this impact functionality if you override |
Tested in the case of CP=1 this is fine. Not confident about CP>1. Do you have suggestions on how to tackle CP>1? |
I am thinking shouldn't we fix |
Yes, I tested cu_seqlens_unpadded != cu_seqlens with TE < 1.13 and it works. I don't think I used TE 1.13 at all. |
nemo/collections/nlp/models/language_modeling/megatron_gpt_model.py
Outdated
Show resolved
Hide resolved
Signed-off-by: Chen Cui <[email protected]>
Certainly. The current PR just fixes a bug in nemo 1. We will propagate the change to nemo 2 in a future PR. |
Signed-off-by: Chen Cui <[email protected]>
Signed-off-by: cuichenx <[email protected]>
beep boop 🤖: 🙏 The following files have warnings. In case you are familiar with these, please try helping us to improve the code base. Your code was analyzed with PyLint. The following annotations have been identified:
Mitigation guide:
By applying these rules, we reduce the occurance of this message in future. Thank you for improving NeMo's documentation! |
[🤖]: Hi @cuichenx 👋, We wanted to let you know that a CICD pipeline for this PR just finished successfully So it might be time to merge this PR or get some approvals I'm just a bot so I'll leave it you what to do next. //cc @pablo-garay @ko3n1g |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. Thanks!
* fix nemo 1 packed sequence TE version error Signed-off-by: Chen Cui <[email protected]> * limit condition to only CP=1 case Signed-off-by: Chen Cui <[email protected]> * add cudnn version guard Signed-off-by: Chen Cui <[email protected]> * Apply isort and black reformatting Signed-off-by: cuichenx <[email protected]> --------- Signed-off-by: Chen Cui <[email protected]> Signed-off-by: cuichenx <[email protected]> Co-authored-by: cuichenx <[email protected]> Signed-off-by: Abhinav Garg <[email protected]>
* fix nemo 1 packed sequence TE version error Signed-off-by: Chen Cui <[email protected]> * limit condition to only CP=1 case Signed-off-by: Chen Cui <[email protected]> * add cudnn version guard Signed-off-by: Chen Cui <[email protected]> * Apply isort and black reformatting Signed-off-by: cuichenx <[email protected]> --------- Signed-off-by: Chen Cui <[email protected]> Signed-off-by: cuichenx <[email protected]> Co-authored-by: cuichenx <[email protected]> Signed-off-by: Youngeun Kwon <[email protected]>
What does this PR do ?
Adds a te version guard for packed sequence in NeMo 1 (NeMo 2 is unaffected)
Changes in #10688 should be paired with TE version >1.3
https://github.com/NVIDIA/TransformerEngine/blob/3d63cbb469ace2d1ad7798a956e16dee42bd655b/transformer_engine/pytorch/attention.py#L8179-L8185
Collection: [Note which collection this PR will affect]
Changelog
Usage
# Add a code snippet demonstrating how to use this
GitHub Actions CI
The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.
The GitHub Actions CI will run automatically when the "Run CICD" label is added to the PR.
To re-run CI remove and add the label again.
To run CI on an untrusted fork, a NeMo user with write access must first click "Approve and run".
Before your PR is "Ready for review"
Pre checks:
PR Type:
If you haven't finished some of the above items you can still open "Draft" PR.
Who can review?
Anyone in the community is free to review the PR once the checks have passed.
Contributor guidelines contains specific people who can review PRs to various areas.
Additional Information