Question about Table. 2 in the paper

Question

Question about Table. 2 in the paper

vtddggg opened this issue 2 years ago · 2 comments

How can we get the results in above figure. Do we need to design text prompts for each task and use them to init the classification head?

I try to add the classification head with random init weights, but get poor results for WiSE-FT.

Answer 1 · 2022-11-01T16:33:33.000Z

Hi @vtddggg, thanks for the interest in our work! We indeed initialize the classification head using text prompts. I recommend using the prompts from OpenAI (see here).

Answer 2 · 2022-11-02T01:15:53.000Z

Thank for your reply!
It has solved my question. Closed