[ET-VK][ez] Make squeeze insertion requirements more strict #9950
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/9950
Note: Links to docs will display an error until the docs builds have been completed. This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
kirklandsign
approved these changes
Apr 7, 2025
Pull Request resolved: #9917 ## Context Refactor the `SqueezeUnsqueezeInputs` pass to be more clear about its intention. For Llama models, input shapes to 4 bit linear will oftentimes have the shape `[1, seq_len, dim]`; under the current implementation of the pass, the input would be squeezed to `[seq_len, dim]` even though the squeeze is not necessary. The original intention of thispass was to squeeze inputs with shape `[batch_size, 1, dim]` to `[batch_size, dim]` before calling the 4-bit linear operator. ## Changes To avoid inserting unnecessary squeeze/unsqueezes, be more specific about when squeeze/unsqueeze should be added. I would like to consider refactoring this pass in the future, since the logic is currently a bit uninttuitive. Squeeze/unsqueeze is also inserted for gelu and relu, but this is to create a chain of unsqueeze/squeeze that will be eliminated by a later pass (see #8601 / D69673068). I think eventually it will be good to rewrite the pass to make shape management more explicit and self contained within the pass rather than inserting ops which are expected to be removed later on. ghstack-source-id: 276566115 @exported-using-ghexport Differential Revision: [D72480178](https://our.internmc.facebook.com/intern/diff/D72480178/)
kirklandsign
force-pushed
the
gh/SS-JIA/207/orig
branch
from
75f61e4 to
e6e6880
Compare
kirklandsign
added
the
release notes: vulkan
kirklandsign
pushed a commit
that referenced
this pull request
Apr 11, 2025
## Context Refactor the `SqueezeUnsqueezeInputs` pass to be more clear about its intention. For Llama models, input shapes to 4 bit linear will oftentimes have the shape `[1, seq_len, dim]`; under the current implementation of the pass, the input would be squeezed to `[seq_len, dim]` even though the squeeze is not necessary. The original intention of thispass was to squeeze inputs with shape `[batch_size, 1, dim]` to `[batch_size, dim]` before calling the 4-bit linear operator. ## Changes To avoid inserting unnecessary squeeze/unsqueezes, be more specific about when squeeze/unsqueeze should be added. I would like to consider refactoring this pass in the future, since the logic is currently a bit uninttuitive. Squeeze/unsqueeze is also inserted for gelu and relu, but this is to create a chain of unsqueeze/squeeze that will be eliminated by a later pass (see #8601 / D69673068). I think eventually it will be good to rewrite the pass to make shape management more explicit and self contained within the pass rather than inserting ops which are expected to be removed later on. Differential Revision: [D72480178](https://our.internmc.facebook.com/intern/diff/D72480178/)
This was referenced Apr 14, 2025
keyprocedure
pushed a commit
to keyprocedure/executorch
that referenced
this pull request
Apr 21, 2025
…9950) ## Context Refactor the `SqueezeUnsqueezeInputs` pass to be more clear about its intention. For Llama models, input shapes to 4 bit linear will oftentimes have the shape `[1, seq_len, dim]`; under the current implementation of the pass, the input would be squeezed to `[seq_len, dim]` even though the squeeze is not necessary. The original intention of thispass was to squeeze inputs with shape `[batch_size, 1, dim]` to `[batch_size, dim]` before calling the 4-bit linear operator. ## Changes To avoid inserting unnecessary squeeze/unsqueezes, be more specific about when squeeze/unsqueeze should be added. I would like to consider refactoring this pass in the future, since the logic is currently a bit uninttuitive. Squeeze/unsqueeze is also inserted for gelu and relu, but this is to create a chain of unsqueeze/squeeze that will be eliminated by a later pass (see pytorch#8601 / D69673068). I think eventually it will be good to rewrite the pass to make shape management more explicit and self contained within the pass rather than inserting ops which are expected to be removed later on. Differential Revision: [D72480178](https://our.internmc.facebook.com/intern/diff/D72480178/)
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 was created by the merge bot to help merge the original PR into the main branch.
ghstack PR number: #9917 by @SS-JIA
^ Please use this as the source of truth for the PR details, comments, and reviews
ghstack PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/207/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/207/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/206/orig
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/207/orig
@diff-train-skip-merge