[BUG] Chapter 10: number of output arguments of model.evaluate(...) for multiple outputs

[thomas-haslwanter]

Describe the bug
cell 81 in "10_neural_nets_with_keras.ipynb" brings up an error

To Reproduce

weighted_sum_of_losses, main_loss, aux_loss, main_rmse, aux_rmse = eval_results

And if you got an exception, please copy the full stacktrace here:

ValueError                                Traceback (most recent call last)
Cell In[64], line 2
      1 eval_results = model.evaluate((X_test_wide, X_test_deep), (y_test, y_test))
----> 2 weighted_sum_of_losses, main_loss, aux_loss, main_rmse, aux_rmse = eval_results

ValueError: not enough values to unpack (expected 5, got 3)

Expected behavior
No errormessage

Versions (please complete the following information):

• OS: Win 11 Education, Build 10.0.22631
• Python: 3.12.2
• TensorFlow: 2.16.1

Additional Info
4 cells below, the same error occurs. In addition
model.compile(loss="mse", loss_weights=[0.9, 0.1], optimizer=optimizer, metrics=["RootMeanSquaredError"])
has to be replaced with
model.compile(loss="mse", loss_weights=[0.9, 0.1], optimizer=optimizer, metrics=("RootMeanSquaredError", "RootMeanSquaredError"))
Otherwise, the following error message appears

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[74], line 7
      5 model.norm_layer_wide.adapt(X_train_wide)
      6 model.norm_layer_deep.adapt(X_train_deep)
----> 7 history = model.fit(
      8     (X_train_wide, X_train_deep), (y_train, y_train), epochs=10,
      9     validation_data=((X_valid_wide, X_valid_deep), (y_valid, y_valid)))
     10 eval_results = model.evaluate((X_test_wide, X_test_deep), (y_test, y_test))
     11 weighted_sum_of_losses, main_loss, aux_loss, main_rmse, aux_rmse = eval_results

File C:\Programs\WPy64-31220\python-3.12.2.amd64\Lib\site-packages\keras\src\utils\traceback_utils.py:122, in filter_traceback.<locals>.error_handler(*args, **kwargs)
    119     filtered_tb = _process_traceback_frames(e.__traceback__)
    120     # To get the full stack trace, call:
    121     # `keras.config.disable_traceback_filtering()`
--> 122     raise e.with_traceback(filtered_tb) from None
    123 finally:
    124     del filtered_tb

File C:\Programs\WPy64-31220\python-3.12.2.amd64\Lib\site-packages\keras\src\trainers\compile_utils.py:250, in CompileMetrics._build_metrics_set(self, metrics, num_outputs, output_names, y_true, y_pred, argument_name)
    248 if isinstance(metrics, (list, tuple)):
    249     if len(metrics) != len(y_pred):
--> 250         raise ValueError(
    251             "For a model with multiple outputs, "
    252             f"when providing the `{argument_name}` argument as a "
    253             "list, it should have as many entries as the model has "
    254             f"outputs. Received:\n{argument_name}={metrics}\nof "
    255             f"length {len(metrics)} whereas the model has "
    256             f"{len(y_pred)} outputs."
    257         )
    258     for idx, (mls, yt, yp) in enumerate(
    259         zip(metrics, y_true, y_pred)
    260     ):
    261         if not isinstance(mls, list):

ValueError: For a model with multiple outputs, when providing the `metrics` argument as a list, it should have as many entries as the model has outputs. Received:
metrics=['RootMeanSquaredError']
of length 1 whereas the model has 2 outputs.

[ItsMacto]

I found a fix to the error. It is due to the Keras version. Google colab uses Keras version: 3.4.1. After some digging I had 3.6.0 locally. This was causing the bug.

To fix the bug run Keras version: 3.4.1
pip install keras==3.4.1