Currently, the gpus argument from the CLI in our training examples doesn't get passed to model.predict_as_dataframe, essentially forcing the prediction to happen on CPU.
This has caused some confusion for users that have tried to extend the training examples into training scripts for their own use cases.
This PR passes the argument to the model.predict_as_dataframe.
Currently, the
gpus
argument from the CLI in our training examples doesn't get passed tomodel.predict_as_dataframe
, essentially forcing the prediction to happen on CPU.This has caused some confusion for users that have tried to extend the training examples into training scripts for their own use cases.
This PR passes the argument to the
model.predict_as_dataframe
.