Regression test: keras deserialization error: 'keras.metrics.mse'

[fbrausse]

I can't run regression test no. 16 on the master branch (command line as generated by smlp_regr.py):

~/intel/smlp-pip/regr_smlp/code $ ../../src/run_smlp.py -model_name "../models/Test8_smlp_toy_num_resp_mult" -out_dir ./ -pref Test16 -mode predict -resp y1,y2 -feat x,p1,p2 -model nn_keras -save_model f -use_model t -mrmr_pred 0 -plots f -pred_plots f -resp_plots f -seed 10  -log_time f  -new_dat "../data/smlp_toy_num_resp_mult_pred_labeled.csv"
[...]
smlp_logger - INFO - Seving model rerun configuration in file ./../models/Test8_smlp_toy_num_resp_mult_rerun_model_config.json
Traceback (most recent call last):
  File "/home/kane/intel/smlp-pip/regr_smlp/code/../../src/run_smlp.py", line 17, in <module>
    main(sys.argv)
  File "/home/kane/intel/smlp-pip/regr_smlp/code/../../src/run_smlp.py", line 13, in main
    smlpInst.smlp_flow()
  File "/home/kane/intel/smlp-pip/src/smlp_py/smlp_flows.py", line 316, in smlp_flow
    model = self.modelInst.build_models(args.model, X, y, X_train, y_train, X_test, y_test, X_new, y_new,
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/intel/smlp-pip/src/smlp_py/smlp_models.py", line 442, in build_models
    model = keras_load_model(self.model_filename(algo, '.h5'))
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/saving/saving_api.py", line 196, in load_model
    return legacy_h5_format.load_model_from_hdf5(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/legacy/saving/legacy_h5_format.py", line 159, in load_model_from_hdf5
    **saving_utils.compile_args_from_training_config(
      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/legacy/saving/saving_utils.py", line 143, in compile_args_from_training_config
    loss = _deserialize_nested_config(losses.deserialize, loss_config)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/legacy/saving/saving_utils.py", line 204, in _deserialize_nested_config
    return {
           ^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/legacy/saving/saving_utils.py", line 205, in <dictcomp>
    k: _deserialize_nested_config(deserialize_fn, v)
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/legacy/saving/saving_utils.py", line 202, in _deserialize_nested_config
    return deserialize_fn(config)
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/losses/__init__.py", line 155, in deserialize
    return serialization_lib.deserialize_keras_object(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/saving/serialization_lib.py", line 590, in deserialize_keras_object
    return deserialize_keras_object(
           ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/saving/serialization_lib.py", line 693, in deserialize_keras_object
    return _retrieve_class_or_fn(
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/home/kane/.local/lib/python3.11/site-packages/keras/src/saving/serialization_lib.py", line 825, in _retrieve_class_or_fn
    raise ValueError(
ValueError: Could not deserialize 'keras.metrics.mse' because it is not a KerasSaveable subclass

I've tried keras versions 3.11.3, 3.12.0 and the latest 3.13.2. All produce the same error.