Summary While attempting to install Keras in editable mode and execute its test suite, several tests failed. Below is a detailed account of the steps taken and the errors encountered.

Steps to Reproduce Created a Conda environment:

bash Copy Edit conda create -n keras python=3.10
conda activate keras
Cloned the Keras repository:

bash Copy Edit git clone https://github.com/keras-team/keras.git
cd keras
Checked out the latest release:

bash Copy Edit git fetch --tags
git checkout tags/v3.8.0
Installed the requirements:

bash Copy Edit python3.10 -m pip install -r requirements.txt
./shell/api_gen.sh
Installed Keras in editable mode:

bash Copy Edit python3.10 -m pip install -e .
Configured the backend:

bash Copy Edit export KERAS_BACKEND="tensorflow"
Ran the test suite:

bash Copy Edit python3.10 -u -m pytest --cov=keras --cov-report=html:htmlcov --cov-report=xml:coverage.xml > keras_golden_run
Observed Behavior The test suite reported the following summary of errors:

Integration Test Failures jax_custom_fit_test.py::test_custom_fit

Error: TypeError CustomModel.train_step() missing 1 required positional argument: 'data' tf_distribute_training_test.py::test_model_fit

Error: RuntimeError Virtual devices cannot be modified after being initialized torch_custom_fit_test.py::test_custom_fit

Error: AttributeError 'CustomModel' object has no attribute 'zero_grad' torch_workflow_test.py::TorchWorkflowTest::test_keras_layer_in_nn_module

Error: AssertionError 0 != 2 Application Test Failures (EfficientNetV2) Several tests failed with InvalidArgumentError, involving the inability to broadcast tensor shapes:

Examples: applications_test.py::test_application_notop_custom_input_shape_EfficientNetV2B0_channels_first applications_test.py::test_application_pooling_EfficientNetV2B3_channels_first Error Message: Unable to broadcast tensor of shape [3] to tensor of shape [1,3,1,1] [Op:BroadcastTo] Summary of Test Results 12 failed, 12870 passed, 346 skipped, 8 xfailed, 2 xpassed Total time: ~21 minutes Environment Details Python Version: 3.10 Anaconda Version: 4.10.1 OS: Ubuntu 22.04.5 LTS Additional Notes Editable Mode: Used pip install -e . instead of pip_build.py to install Keras. Potential Causes: Integration test failures may indicate issues with the Jax and PyTorch configurations. Application test errors might involve TensorFlow device setups or input shape mismatches for EfficientNetV2 models. Request for Assistance I expected the test suite to pass or, at the very least, have no critical errors.

Are these test failures indicative of underlying issues that require resolution? Should I reconfigure the backend to include Jax and PyTorch for these tests? Are additional dependencies or configurations needed for a successful test run? If TensorFlow is my primary backend, can I safely ignore failures related to Jax and PyTorch? Any guidance on addressing these issues would be greatly appreciated.

Thank you!

Comment From: sonali-kumari1

Hi @ARforyou -

Thanks for reporting this. Please refer to this similar issue.

Comment From: github-actions[bot]

This issue is stale because it has been open for 14 days with no activity. It will be closed if no further activity occurs. Thank you.

Comment From: github-actions[bot]

This issue was closed because it has been inactive for 28 days. Please reopen if you'd like to work on this further.