Question about text prototype in reprogramming

Question

Question about text prototype in reprogramming

Closed this issue 5 months ago · 3 comments

Hi, there. Thanks for publishing your code. I'm interested in your patch reprogramming. But I didn't find your text prototype in your code TimeLLM.py

This is only the linear projection. I'm not sure if I misunderstand the code. Thanks for your reply.

Answer 1 · 2024-05-02T02:00:59.000Z

It seems like you've caught on quite well. The authors mentioned in the paper, "A simple solution is to maintain a small collection of text prototypes by linearly probing E, denoted as E'." The part you mentioned about the mapping_layer likely corresponds to this.

Answer 2 · 2024-05-06T14:39:49.000Z

I think they import the whole model word-embedding weight and then get the text prototype by this linear layer("self.mapping_layer").

Answer 3 · 2024-05-15T04:26:41.000Z

I think they import the whole model word-embedding weight and then get the text prototype by this linear layer("self.mapping_layer").

Yes, your understanding is correct. We have also provided a detailed description in the "Patch Reprogramming" section of our paper, which you can refer to.