Issues when implementing 'predict_model'

[tueseu]

Thank you for the instruction about integrating Pytorch models into the Pycaret workflow. The example you provide works well. But when I go further to implement the developed skorch_model for prediction, an error arises. Do you have any idea how to fix it?

The error is shown as follows:

predict = predict_model(skorch_model, data)

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-16-74e96e3a1dc7> in <module>
----> 1 predict = predict_model(skorch_model, data)

C:\Program Files\Python38\lib\site-packages\pycaret\classification.py in predict_model(estimator, data, probability_threshold, encoded_labels, raw_score, round, verbose)
   2003     """
   2004 
-> 2005     return pycaret.internal.tabular.predict_model(
   2006         estimator=estimator,
   2007         data=data,

C:\Program Files\Python38\lib\site-packages\pycaret\internal\tabular.py in predict_model(estimator, data, probability_threshold, encoded_labels, raw_score, round, verbose, ml_usecase, display)
   8712 
   8713         if is_sklearn_pipeline(estimator) and hasattr(estimator, "predict"):
-> 8714             dtypes = estimator.named_steps["dtypes"]
   8715         else:
   8716             try:

KeyError: 'dtypes'

Thank you for your kind help!

[pranaymodukuru]

Hi @tueseu is the variable data here a dataframe?