[WIP] Fetch new validation data for each calc_loss #279
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
Sopel97
changed the title
|
As I already said in discord: the loss now really jumps around like crazy. My idea was to only slowly introduce new positions to the validation set to get a more stable loss output which will help with automatic LR drops. |
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Outdated suggestions cannot be applied.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.
This PR makes
calc_lossuse a new validation set each time it's called (one can now limit the amount of validation data used percalc_losscall, so this makes sense). Since we're using relatively small validation sets for each calc_loss this should, in principle at least, make validation loss more informative when observed over multiple validation steps.I'd like to get some feedback from @noobpwnftw on this change before it's merged.
I'd also like to get some feedback on how to handle the case when we couldn't fetch the requested amount of the data for validation. This might happen in extreme cases. Currently I made it continue with a warning as the training can proceed even without validation data, but there might be a better resolution to this problem.