fix tf.nn.{conv2d,convolution} substitution #1275
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
elad-c
reviewed
Nov 24, 2024
| if b is None: | ||
| conv_fw_attr[USE_BIAS] = False | ||
| else: | ||
| weights[BIAS] = b | ||
|
|
||
| data_format = conv_func_node.op_call_kwargs.get(DATA_FORMAT, 'NHWC') | ||
| conv_fw_attr['data_format'] = {'NHWC': 'channels_last', 'NCHW': 'channels_first'}[data_format] |
There was a problem hiding this comment.
use constants from the keras constants
| """ | ||
| v = node.op_call_kwargs.get(key) | ||
| if v is None: | ||
| return 1, 1 |
There was a problem hiding this comment.
this way you assume the defaults. why not return None?
There was a problem hiding this comment.
That's intentional. None wouldn't do, we need to fill in an explicit default. This method is specific to tf stride & dilation
| @@ -54,6 +54,8 @@ def compare(self, quantized_model, float_model, input_x=None, quantization_info= | |||
| y = float_model.predict(input_x) | |||
| y_hat = quantized_model.predict(input_x) | |||
| self.unit_test.assertTrue(y.shape == y_hat.shape, msg=f'out shape is not as expected!') | |||
| # FIXME this doesn't test anything, the number of quantized convs in the network is exactly 0. Even if it | |||
| self.unit_test.assertTrue(len(layer.weights) == 2,msg=f'fail Bias should appear in weights!!') | ||
|
|
||
|
|
||
| class FuncConv2DCollapsingTest(FourConv2DCollapsingTest): |
There was a problem hiding this comment.
These tests seem redunant, as the there's no difference between these tests and the ones for Conv2D layer. wht not just test the substitution of these layers to Conv2D?
irenaby
changed the title
elad-c
approved these changes
Nov 24, 2024
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.
Pull Request Description:
Fixed tf conv substitution to handle attrs with default values that were not passed explicitly, and convert tf-specific format to keras compatible values.
Checklist before requesting a review: