float_seq without discretization?

[ElkanaBaris]

Hi,

I'm trying to add an input of float_seq, some external embedding, to my model.
It seems that although I didn't configure it, the embedding undergo a discretization process.
I don't want to embed it again.
Is that true?
What is the thought process behind it?
How can I disable this functionality?

Thanks!

BTW - you have an awesome repo!

P.S. attached is the code from recbole/data/dataset/dataset.py:

[kerengaiger]

I also struggle with that issue