I’m trying to use the use_nn argument of collab_learner but it keeps giving errors. For instance, the command:
collab_learner(data, n_factors=50, use_nn = True, metrics= root_mean_squared_error, y_range=y_range, wd=1e-2)
Returns an error saying metrics isn’t a valid argument. I remove the metrics and it returns the samething for weight decay argument (wd).
Anybody had this problem?
I know this topic is old, but I ran into the same problem and I think it is due to an error in the package source.
collab_learner creates a new EmbedddingNN object with a **learn_kwargs argument. But the **learn_kwargs should be passed only to the CollabLearner class. If you rewrite the collab_learner function without passing the **learn_kwargs when the EmbedddingNN is created, everything seems to work fine.