Extract category labels for my embedding

Part 1 (2019)
1

I’m using the tabular learner and want to extract the embedding matrices along with the category labels.

To extract the embedding matrix for my first categorical variable, I use the following:
to_np(learn.model.embeds[0].weight)

This works fine and gives me a matrix of dimensions 10x6, as expected. I now would like to get the labels for the 10 rows in this embedding but can’t seem to find anything on how to extract them in v1 of the library.

Thanks

2

You should look at my thread about transfer learning, its pretty easy to do.

1 Like