fix R_Square shape issue in model.evaluate

[schmidt-jake]

Description

The R_Square metric was broken when calling model.evaluate or model.fit with validation_data set. This was due to a tensor shape mismatch.

Fixes # (issue)

Type of change

• Bug fix
• New Tutorial
• Updated or additional documentation
• Additional Testing
• New Activation (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Callback (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Image addition (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Layer (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Loss (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Metric (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Optimizer (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New RNN Cell (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Seq2seq addition (Requires creating an issue first)
  • The changes conform to the contribution guidelines
• New Text addition (Requires creating an issue first)
  • The changes conform to the contribution guidelines

How Has This Been Tested?

If you're adding a bugfix or new feature please describe the tests that you ran to verify your changes:

• I added to metrics/test/r_square_test.py:

def test_keras_fit():
    model = tf.keras.Sequential([tf.keras.layers.Dense(1)])
    model.compile(loss="mse", metrics=[RSquare(y_shape=(1,))])
    data = tf.data.Dataset.from_tensor_slices(
        (tf.random.normal(shape=(100, 1)), tf.random.normal(shape=(100, 1)))
    )
    data = data.batch(10)
    model.fit(x=data)
    model.fit(x=data, validation_data=data)

Checklist:

• I've properly formatted my code according to the guidelines
  • By running Black + Flake8
  • By running pre-commit hooks
• This PR addresses an already submitted issue for TensorFlow Addons
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• This PR contains modifications to C++ custom-ops

[bot-of-gabrieldemarmiesse]

@marload @SSaishruthi

You are owners of some files modified in this pull request.
Would you kindly review the changes whenever you have the time to?
Thank you very much.

[bhack]

Thanks. See the first point on the checklist.

[archsyscall]

Thank You!

Could you solve the test failure?

First of all, flake8 will be solved by removing the import of weights_broadcast_ops, which is an unused function.
Also, the test_r2_sklearn_comparison has an error.

[schmidt-jake]

@marload I believe I made all the necessary changes. All the tests are passing except one build... is this my fault? Please advise.

[archsyscall]

The failed test seems to fail in all other PRs. I think you'd better write an Issue.

[WindQAQ]

Thanks 😄