Which prompt was used in the paper's result?

[Hyfred]

Thanks for your nice work, I have a short question:

I noticed both prompt mechanisms(prefix_prompt and suffix_prompt) were used in all three tasks. So my question is which one is the result shown in the paper?

[yyy-Apple]

The results shown in our paper is the avg of all prompts (see section 4.2.2 Settings -- Selection of Prompts for more details). Specifically, we get the score for one hypothesis by taking the average of all its generation scores using different prompts.

[Hyfred]

There are two methods for adding a prompt during score calculation: either appending it to the source text or prepending it to the target text, as described in Section 3.3.1.

Based on the description in Section 4.2.2 and Appendix A.4, is the experimental result derived from conducting the experiment by prefixing the decoder (prepending the prompt to the target text)?

Because I noticed there is another option (prefixing in the source text) as well, it is confusing to me.

[yyy-Apple]

I'm not exactly sure, since it's been a long time. You can check the numbers with analysis.ipynb to see which one matches the results. As I recall, we may have used the avg of prepending to the target text.

[Hyfred]

Got it, thanks for your reply!