Tabular model with only continuous or categorical variables

Is it possible to create such a model?
Just passing cat_names = [] does not seem to work, gives an error at the learner.

It should work - see this recent thread with only one continuous variable and cat_names=[]

Yes thanks, the trick is setting emb-szs={}.
Anyway, data.show_batch(rows=10)does not seem to work.
TypeError: len() of a 0-d tensor

1 Like

Check this.

Lol, @sgugger saving the day!