NorMuon + Deeper U-Net with INT6 Fake Quantization#2016
Open
sea-rod wants to merge 1 commit into
Open
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Swapped out Muon for NorMuon so neurons get more balanced updates instead of a few dominating the whole run. Bumped layers from 9 to 12 for more depth and better skip connections without blowing the size budget. Added INT6 fake quant on attention and MLP activations so the model actually trains closer to how it gets exported, rather than seeing clean fp32 the whole time then getting hit with int8. Also dropped q_gain init from 1.5 to 1.0 since starting inflated after QK-norm was just causing early spikes for no reason.