Hi , Run # 22400 . Error : " There is no trained model to use predict() " . I have a proper train function that causes the code to run 11+ hours , and when it proceeds to the infer function , this is the error I received ! As in supervised notebook , the last part including infer is the same . Nothing in the infer and submission has changed from your supervised notebook, only the training uses more complex methodology and various features.
It is hard to debug your code without being able to see it.
Have you perhaps accidentally created a new (so empty) model instance and forgotten to:
- load it from disk?
- train it with the data provided?