[NNAPI] Handle binary ops combining NHWC+NCHW in some cases #48812
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
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. [ghstack-poisoned]
This was referenced Dec 3, 2020
Closed
Closed
Closed
dreiss
added a commit
that referenced
this pull request
Dec 3, 2020
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. ghstack-source-id: 6e67e44e59e6386504d28db0b78d9dcc395e9154 Pull Request resolved: #48812
💊 CI failures summary and remediationsAs of commit ccd0856 (more details on the Dr. CI page):
This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions to the (internal) Dr. CI Users group. |
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
|
Looks like this PR hasn't been updated in a while so we're going to go ahead and mark this as |
axitkhurana
approved these changes
Feb 10, 2021
|
|
||
| out_oper = oper._replace(dim_order=DimOrder.CHANNELS_LAST) | ||
|
|
||
| inputs = [None] * 2 |
There was a problem hiding this comment.
any reason to create a [None, None] list and then replacing values vs just creating inputs = [in_id, self.add_immediate_int_vector([0, 2, 3, 1])?
(same with outputs)
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
This was referenced Mar 25, 2021
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
Summary: This came up in a squeeze-and-excitation model. Starting with an NHWC tensor T, we perform a mean operation across H and W, giving an NxC tensor, which (after some fully connected layers) is reshaped to NxCx1x1, then multiplied with T. To handle this, we detect the specific case of a binary op with one NHWC input and one contiguous input with H,W == 1,1 and allow the op to be applied (after transposing the contiguous input). Test Plan: Unit test. Differential Revision: [D25317939](https://our.internmc.facebook.com/intern/diff/D25317939) [ghstack-poisoned]
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.
Stack from ghstack:
Summary:
This came up in a squeeze-and-excitation model. Starting with an NHWC
tensor T, we perform a mean operation across H and W, giving an NxC
tensor, which (after some fully connected layers) is reshaped to
NxCx1x1, then multiplied with T. To handle this, we detect the specific
case of a binary op with one NHWC input and one contiguous input with
H,W == 1,1 and allow the op to be applied (after transposing the
contiguous input).
Test Plan:
Unit test.
Differential Revision: D25317939