fix: accumulate train_loss correctly across micro-steps#357
Open
sjhddh wants to merge 1 commit intokarpathy:masterfrom
Open
fix: accumulate train_loss correctly across micro-steps#357sjhddh wants to merge 1 commit intokarpathy:masterfrom
train_loss correctly across micro-steps#357sjhddh wants to merge 1 commit intokarpathy:masterfrom
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Currently,
train_loss = loss.detach()overwrites the loss tracking on each micro-step, meaning the reportedtrain_loss_fsolely reflects the loss of the final micro-batch instead of the true mean of the entire global batch.This PR fixes this by accumulating
loss.detach() / grad_accum_stepsover the loop. This reduces the variance in the loggeddebiased_smooth_lossand provides an exact mean loss corresponding to the global batch.