How do I load model after the training is complete ?

fit method is being invoked every time a call is made and job results were successful but when the next prediction happens self.train_output is always blank. Any pointers on how to get self.train_output

Start by tracing the backend's fit and prediction entry points, focusing on when self.train_output is assigned and when it is read. Reproduce the successful training followed by a prediction, then determine where the trained result is lost; done means the next prediction can access the completed training output.

Written by the indexing model from the issue text.