Feat/gradient statistics monitor callback#21720
Open
JanFidor wants to merge 8 commits into
Open
Conversation
JanFidor
requested review from
ethanwharris,
justusschock and
tchaton
as code owners
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Closes #21589
Credits to @ZiadiSafouene whose original PR #21596 this one was heavily based on.
This PR provides implementation of GradientStatsMonitor callback for Lightning that tracks gradient statistics
during training.
The callbacks allows the user to track both global and per layer gradient statistics as well as per epoch and per batch ones.
Implementation should make extending this class simple with functions like compute_batch_stats or per epoch statistics functions following conventions from torchmetrics: initialize, update, compute.
The callback supports state_dicts -> interrupting and resuming training.
Provides easy diagnostics for high gradients even when working with gradient clipping by utilizing on_before_optimizer_step hook.
Related tests are included.
In short this callback provides a simple out of the box solution for tracking ML model gradients.
Before submitting
PR review
Anyone in the community is welcome to review the PR.
Before you start reviewing, make sure you have read the review guidelines. In short, see the following bullet-list:
Reviewer checklist
📚 Documentation preview 📚: https://pytorch-lightning--21720.org.readthedocs.build/en/21720/