Collaborative filtering - How to predict?

With regards to the movie collaborative filtering mode. How do I make a prediction for a specific user and movie?
For instance I did this to get the score for user 460 and movie FLy away home 1996

new_df = pd.DataFrame({‘user’:[460],‘title’:[‘Fly Away Home (1996)’]})
dl = learn.dls.test_dl(new_df)

But I’m getting the following error
IndexError: too many indices for tensor of dimension 0