Skip to content

Conversation

@jikechao
Copy link
Member

@jikechao jikechao commented Jun 8, 2023

This PR fixed two bugs in Conv2DTranspose and Conv3DTranspose. Both of them are caused by lacking the parsing of the output_padding attribute.
In addition, corresponding test cases were added to capture the bug.

Steps to reproduce

import tvm
import tvm.relay as relay
import numpy as np
from tensorflow import keras
from tensorflow.keras import layers, models
input_shape = (2, 33,33,128)
input_data = np.random.random(size=input_shape)
x = layers.Input(shape=input_shape[1:], dtype='float32')

kwargs={'filters':2,'kernel_size':[3, 3],'strides':[2, 2],'output_padding':[1, 1],}# 'data_format':"channels_last"} 
# output_padding lead to this bug
layer = keras.layers.Conv2DTranspose(**kwargs)
layer.set_weights(layer.get_weights())

y = layer(x)
model = models.Model(x, y)
model.summary()
res_keras = model(input_data)

shape_dict = {'input_1': input_shape}
mod, params = relay.frontend.from_keras(model, shape_dict,layout='NHWC')
with tvm.transform.PassContext(opt_level=3):
    model = relay.build_module.create_executor("graph", mod, tvm.cpu(0), 'llvm', params).evaluate()

test_x_tvm = input_data
res_tvm = model(tvm.nd.array(test_x_tvm.astype('float32'))).numpy()

np.testing.assert_allclose(res_keras, res_tvm, atol=1e-3, rtol=1e-3)

image

BTW, The Conv3DTranspose has a similar bug. The following keras layer can trigger it:

keras_mod.layers.Conv3DTranspose(
                filters=2, kernel_size=(3, 3, 3), strides=(2, 2, 2), output_padding=(1, 1, 1)
            ),

cc @echuraev @Hzfengsy @masahi

@tvm-bot
Copy link
Collaborator

tvm-bot commented Jun 8, 2023

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

Generated by tvm-bot

@github-actions github-actions bot requested review from Hzfengsy, echuraev and masahi June 8, 2023 03:50
@masahi masahi merged commit 9e265ff into apache:main Jun 8, 2023
junrushao pushed a commit to junrushao/tvm that referenced this pull request Jun 22, 2023
…g' in Deconv (apache#15060)

* fix deconv about the parameter output_padding

* add test cases to capture the in deconv

* Update keras.py
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants