NXP backend: Add support for depthwise and separable convolution. #11215
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/11215
Note: Links to docs will display an error until the docs builds have been completed. ❌ 3 New FailuresAs of commit 07f79c2 with merge base 77f16dc ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@pytorchbot label "module: nxp" "release notes: nxp" |
| @@ -204,3 +210,328 @@ def test_conv2d_quant_conversion(mocker, model: torch.nn.Module, input_shape): | |||
| input_data=input_data, | |||
| atol=1.0, | |||
| ) | |||
|
|
|||
|
|
|||
| @pytest.mark.parametrize("stride", [1, 2]) | |||
There was a problem hiding this comment.
I thought we decided to not use pytest features in new tests :P
There was a problem hiding this comment.
Sorry for misunderstanding. I thought we are OK using pytest temporarily for the those tests, we already have in our development tree.
The latest (in our development tree) uses the unittest instead of pytest - nxp-upstream@63fd43f#diff-d103a8083dbc6da9184c269a015199baa963f734cdd8cb2f29336e6954f01cb8R7
| assert len(nodes) == 11 | ||
| assert ( | ||
| nodes[7].target.__name__ == "aten.convolution.default" | ||
| ) # Convolution not delegated. |
|
Just skimmed, let me take a close look in a day or so, esp |
The `group` attribute of the `aten.convolution` has an effect on how the weights are used in the computation of the convolution. To properly utilize Neutron, the convolution is sometimes converted to `DepthwiseConv2D`.
8e64288 to
cb91593
Compare
|
Ready for review. Most of the tests passing. The failing seems unrelated to the changes:
|
|
digant is out for a while I will try to find another person to review these |
Neutron only supports batch == 1.
|
Converting to draft unless the NXP Backend CI is back (#11756) |
I am back, I can review this coming week. |
Summary
Adding depthwise convolution support for the NXP backend
Test plan
Functionality tested by python unit tests:
cc @digantdesai @JakeStevens , @Pop-korn