Needs explaination for 'prompt_y=norm(prompt_emb)+prior_y'

[ztb-35]

Dear authors:
Thanks for sharing the code of your great work.
I find that you try to add prior_y to the prompt in the code 'exp/exp_long_term_forecasting.py #578, prompt_y=norm(prompt_emb)+prior_y'.
I guess the reason is that you want to merge the text with time series data. But the authors didn't mention it in their paper.
Any reply is appreciated!

[blacksnail789521]

I found out that "prior_history_avg" is just the old value in "OT".

[ztb-35]

Thanks for your reply.
I think the prior_history_avg is the values in the column titled "prior_history_avg" in the data.
I just wonder why they add the prior_y to the normed text embedding.
The author answers me the question elsewhere. He told me they want to align the text embedding to the numerical, thus they add the prior_y to the normed text embedding here.

[Waguy02]

Author @haoxin1998 , could provide clear explanations please?
Thanks