support for text features and multi-value category features and sequence features

[Lapis-Hong]

1. text features:
   use text encoder as an embedding input, can join with continuous feature to feed to existed models

2. multi-value category features
   like [a, b, c], can process similar to categorical features, but with pooling to sum its embeddings

3. sequence features
   like [s1,s2,...sn], can use embedding an attention

[manujosephv]

I feel these are more nuanced use-cases and form the question, I assume you are trying to sue this for recommendation systems?

PyTorch Tabular is more geared towards standard tabular problems like classification and regression. That is why multi-hot categorical features aren't supported out of the box.

Adding support for Text and Sequence Features is a good idea and I'm all for it. It should be easy enough to include and parametrize that. As things stand now, you can implement custom models to include such features..

[Lapis-Hong]

thanks, i am doing classification and regression probs, in real world big scale data, actually there always exists multi-value data, looking forward for text and sequence features support~

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.