Fixed NotImplementedError: Cannot convert a symbolic Tensor to a numpy array

[petronny]

Fixed #9706

Implemented as mentioned in tensorflow/models#9706 (comment) and tensorflow/models#9706 (comment) .

This problem happens with tensorflow 2.4.1 and numpy 1.20.1 .
Minimal code to reproduce the problem:

from tensorflow.keras.models import Sequential
from tensorflow.keras import layers
model = Sequential()
model.add(layers.LSTM(256, return_sequences=True))
model.build((10,20,30))
model.summary()

[google-cla]

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed (or fixed any issues), please reply here with @googlebot I signed it! and we'll verify it.

---

What to do if you already signed the CLA

Individual signers

• It's possible we don't have your GitHub username or you're using a different email address on your commit. Check your existing CLA data and verify that your email is set on your git commits.

Corporate signers

• Your company has a Point of Contact who decides which employees are authorized to participate. Ask your POC to be added to the group of authorized contributors. If you don't know who your Point of Contact is, direct the Google project maintainer to go/cla#troubleshoot (Public version).
• The email used to register you as an authorized contributor must be the email used for the Git commit. Check your existing CLA data and verify that your email is set on your git commits.
• The email used to register you as an authorized contributor must also be attached to your GitHub account.

ℹ️ Googlers: Go here for more info.

[petronny]

@googlebot I signed it!

[gbaned]

@petronny Can you please check build failures. Thanks!

[petronny]

I think the main error is:

ERROR: testImplementsAttributeAssertsOnSideInput (__main__.FunctionTest)
testImplementsAttributeAssertsOnSideInput (__main__.FunctionTest)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/b/f/w/bazel-out/k8-opt/bin/tensorflow/python/eager/function_test.runfiles/org_tensorflow/tensorflow/python/ops/array_ops.py", line 3219, in ones
    tensor_shape.TensorShape(shape))
  File "/b/f/w/bazel-out/k8-opt/bin/tensorflow/python/eager/function_test.runfiles/org_tensorflow/tensorflow/python/framework/tensor_shape.py", line 762, in __init__
    self._dims = [Dimension(d) for d in dims]
  File "/b/f/w/bazel-out/k8-opt/bin/tensorflow/python/eager/function_test.runfiles/org_tensorflow/tensorflow/python/framework/tensor_shape.py", line 762, in <listcomp>
    self._dims = [Dimension(d) for d in dims]
  File "/b/f/w/bazel-out/k8-opt/bin/tensorflow/python/eager/function_test.runfiles/org_tensorflow/tensorflow/python/framework/tensor_shape.py", line 206, in __init__
    .format(value, type(value))), None)
  File "<string>", line 3, in raise_from
TypeError: Dimension value must be integer or None or have an __index__ method, got value '1.0' with type '<class 'float'>'

I guess it shoule be 1 with type 'int' instead of 1.0 with type 'float'.

Is there a way to easily run the testImplementsAttributeAssertsOnSideInput test on my local machine?
Then I can try to dig this error out.

[mihaimaruseac]

bazel test //path/to/the/test:test_target (need to look in the BUILD files to get the test target name)

[gbaned]

@petronny Can you please check @mihaimaruseac's comments and keep us posted ? Thanks!

[petronny]

I will take a look after 2 weeks.

[csachs]

CI looks better now, I think, can someone take a look?

[risicle]

Is this likely to get a stable backport?