handle static ints and floats #140
Closed
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
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D48667679 |
Summary: Pull Request resolved: pytorch/executorch#140 In order to support MV3 which has decomposed hardswish and hardsigmoid Decomp rules for both: ### Hardswish https://www.internalfb.com/code/fbsource/[9368f8417bd843ee8c91e24ac616ed7f4b194ed8]/xplat/caffe2/torch/_decomp/decompositions.py?lines=182-185 ### Hardsigmoid https://www.internalfb.com/code/fbsource/[9368f8417bd843ee8c91e24ac616ed7f4b194ed8]/xplat/caffe2/torch/_decomp/decompositions.py?lines=159-162 ### Fixing Zero-Dim tensors Both of these decompositions produce zero-dim tensors in the graph ( The + 3 and the / 6). This breaks for XNNPACK because it does not have zero-dim tensors. Instead if the static data is zero dim, then we will interpret it as [1]. #### Fixing torch.int64 static data In the decomposition 3 is converted via to_copy(torch.float32). However 6 remains as an int64. XNNPACK does not handle non-quantized integers, so we also cast all static data that is not quantized to float32 values. Reviewed By: digantdesai Differential Revision: D48667679 fbshipit-source-id: 5edc3d881a599b0f1ee9fc6fddbc582db2f729ee
|
This pull request was exported from Phabricator. Differential Revision: D48667679 |
mcr229
force-pushed
the
export-D48667679
branch
from
fcd4e8b to
332f760
Compare
|
This pull request has been merged in 80ed456. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
Summary:
In order to support MV3 which has decomposed hardswish and hardsigmoid
Decomp rules for both:
Hardswish
https://www.internalfb.com/code/fbsource/[9368f8417bd843ee8c91e24ac616ed7f4b194ed8]/xplat/caffe2/torch/_decomp/decompositions.py?lines=182-185
Hardsigmoid
https://www.internalfb.com/code/fbsource/[9368f8417bd843ee8c91e24ac616ed7f4b194ed8]/xplat/caffe2/torch/_decomp/decompositions.py?lines=159-162
Fixing Zero-Dim tensors
Both of these decompositions produce zero-dim tensors in the graph ( The + 3 and the / 6). This breaks for XNNPACK because it does not have zero-dim tensors. Instead if the static data is zero dim, then we will interpret it as [1].
Fixing torch.int64 static data
In the decomposition 3 is converted via to_copy(torch.float32). However 6 remains as an int64. XNNPACK does not handle non-quantized integers, so we also cast all static data that is not quantized to float32 values.
Reviewed By: digantdesai
Differential Revision: D48667679