-
Notifications
You must be signed in to change notification settings - Fork 3.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Torch] Remove unnecessary reshapes for batch_matmul #7675
Merged
Merged
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
masahi
reviewed
Mar 16, 2021
Pushed a new commit to also reorder the reshape_b and transpose so that the simplify expression can be used. Before this PR:
After this PR:
In particular, since the weights in most PyTorch models have to be transposed when converting to Relay, the second case, for example, could be:
By applying SimplifyExpr to cancel unnecessary
|
masahi
approved these changes
Mar 17, 2021
Thanks @comaniac |
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
May 6, 2021
* [Torch] Remove unnecessary reshapes for batch_matmul * lint * fix * reorder * lint
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
May 11, 2021
* [Torch] Remove unnecessary reshapes for batch_matmul * lint * fix * reorder * lint
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Outdated suggestions cannot be applied.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 removes unnecessary reshape ops in the PyTorch frontend when converting to batch_matmul. This should help the performance of NLP models such as BERT.
cc @siju-samuel @masahi