Add an event related to forward in the TrainerCallback #36496
Conversation
This file contains 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
…er. I placed this event in the compute_loss method in Trainer, and allowed passing inputs and loss to it. The goal is to supply an event related with model forward (to track activation memory and details of inputs). Considered calling the event outside the compute_loss method, everywhere we do forawrd pass on the model, but seemed less generic and more mess.
|
Hi 👋, thank you for opening this pull request! The pull request is converted to draft by default. When it is ready for review, please click the |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
added an event "on_compute_loss" to TrainerCallback and CallbackHandler. I placed this event in the compute_loss method in Trainer, and allowed passing inputs and loss to it. The goal is to supply an event related with model forward (to track activation memory and details of inputs). Considered calling the event outside the compute_loss method, everywhere we do forawrd pass on the model, but seemed less generic and more mess.
What does this PR do?
This PR added an "on_compute_loss" event in the TrainerCallback and Callback handler. In my case - I use this event to log memory metrics so that I can follow activations consumption, but it can also be used to check stats of model inputs, and maybe other things too.
The motivation is discussed here:
#36012
@SunMarc