Create RAI dashboard reports rai_component_utilities.UserConfigError

Question

I registered a mlflow sklearn model into the model registry of AML. Then I came into the Responsible AI tab to create the RAI dashboard. But the job reported an error and failed to create the dashboard.

I pasted the error log below:

Wait 30s for application insights async channel to flush
Traceback (most recent call last):
  File "/mnt/azureml/cr/j/100b64f038e64e8f8b7d0dc31fddebef/exe/wd/rai_component_utilities.py", line 147, in load_mlflow_model
    model = mlflow.pyfunc.load_model(model_uri)._model_impl
  File "/azureml-envs/responsibleai/lib/python3.8/site-packages/mlflow/pyfunc/__init__.py", line 680, in load_model
    model_impl = importlib.import_module(conf[MAIN])._load_pyfunc(data_path)
  File "/azureml-envs/responsibleai/lib/python3.8/site-packages/mlflow/sklearn/__init__.py", line 497, in _load_pyfunc
    _load_model_from_local_file(path=path, serialization_format=serialization_format)
  File "/azureml-envs/responsibleai/lib/python3.8/site-packages/mlflow/sklearn/__init__.py", line 454, in _load_model_from_local_file
    return cloudpickle.load(f)
  File "sklearn/tree/_tree.pyx", line 676, in sklearn.tree._tree.Tree.__setstate__
  File "sklearn/tree/_tree.pyx", line 1364, in sklearn.tree._tree._check_node_ndarray
ValueError: node array from the pickle has an incompatible dtype:
The full logs are attached here: [std_log.txt](/api/attachments/3b3237f2-9583-4390-9c14-b7cc2e1c49d7?platform=QnA)

Full logs:

std_log.txt

Answer 1

Hello Jonas,

Thanks for reaching out to us, I see in your log that the scikit version is not satisfied the requirement -

current: 1.2.2, required: scikit-learn==1.4.2

It seems above is the reason caused the problem. I hope this helps.

Regards,

Yutong

-Please kindly accept the answer if you feel helpful to support the community, thanks a lot.