Replace shape-manipulating subgraphs with ComputeShape operator#1187
Draft
robertknight wants to merge 2 commits intomainfrom
Draft
Replace shape-manipulating subgraphs with ComputeShape operator#1187robertknight wants to merge 2 commits intomainfrom
ComputeShape operator#1187robertknight wants to merge 2 commits intomainfrom
Conversation
fdeceaa to
cd8e3be
Compare
Use symbolic tensor values produced by shape inference to replace subgraphs in models with a `ComputeShape` operator. In the Llama 3 model, this reduces the total number of operators run on each inference pass from 1740 to 1192 (-31%). This reduces interpreter overhead, although the impact is usually small because shape-manipulating subgraphs are operating on small values. More importantly, it removes more potentially fusion-blocking value consumers from the graph. This generalizes the idea of `ComputeShapeFusion`.
cd8e3be to
8761770
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Use symbolic tensor values produced by shape inference to replace subgraphs in models with a
ComputeShapeoperator.In the Llama 3 3B model, this reduces the total number of operators run on each inference pass from 1740 to 1192 (-31%). This reduces interpreter overhead, although the impact is usually small because shape-manipulating subgraphs are
operating on small values. More importantly, it removes more potentially fusion-blocking value consumers from the graph. This generalizes the idea of
ComputeShapeFusion.Fixes #270
TODO:
ComputeShapeto something more general?