WIP: try folding arithmetic forward into polynomial.eval#3046
Open
copybara-service[bot] wants to merge 1 commit into
Open
WIP: try folding arithmetic forward into polynomial.eval#3046copybara-service[bot] wants to merge 1 commit into
copybara-service[bot] wants to merge 1 commit into
Conversation
This idea came out of the matvec + batch_norm fusing: if we can't fuse into the preceding layer, then maybe we can fuse into the next operation. If it's a polynomial.eval op from an activation approximation, then simple ops like those involved in a batch_norm become a polynomial composition, and can be folded into the polynomial.eval op's coefficients. One can also fold addition ops after a polynomial.eval into the eval. This initial implementation was LLM-generated. I still need to do a few things to clean up and validate the approach. PiperOrigin-RevId: 923460715
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
WIP: try folding arithmetic forward into polynomial.eval
This idea came out of the matvec + batch_norm fusing: if we can't fuse into the preceding layer, then maybe we can fuse into the next operation. If it's a polynomial.eval op from an activation approximation, then simple ops like those involved in a batch_norm become a polynomial composition, and can be folded into the polynomial.eval op's coefficients. One can also fold addition ops after a polynomial.eval into the eval.
This initial implementation was LLM-generated. I still need to do a few things to clean up and validate the approach.