LAMB and gradient clipping (instructions vs api)
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 9k
- Forks
- 1.5k
- Avg merge
- 2d 4h
- Merged PRs (30d)
- 3
Description
https://nvidia.github.io/apex/advanced.html?highlight=accumulate#gradient-clipping
The instructions above show how to use pytorch's gradient clipping api for APEX.
However, when using the Fused LAMB optimizer, there is an argument 'max_grad_norm' it suggests that the Fused LAMB optimizer implements gradient clipping inside.
So is there any need for implementing gradient clipping as mentioned in the instructions in the link?
Thank you
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the gradient-clipping section of advanced.html and the Fused LAMB optimizer API, focusing on the max_grad_norm argument. Compare the two documented approaches and clarify whether using max_grad_norm makes the separate PyTorch gradient-clipping step unnecessary.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- documentation, machine-learning
- Issue type
- Documentation
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100