Why concat the `Final Solutions` and `Rollout Solutions` lists in `do_discriminate.py`?

[BaixuanLi]

Hi, I noticed a section of code in do_discriminate.py:

trace_js = read_json(answer_js_file.replace("Answer", "Final Solutions")) + read_json( answer_js_file.replace("Answer", "Rollout Solutions") )

Could you please clarify why the Final Solutions and Rollout Solutions lists are concatenated? Based on my understanding, the Final Solutions list should be a subset of the Rollout Solutions list. Wouldn’t this design introduce redundancy in the evaluation or lead to duplicate judgments? I would greatly appreciate any insights into the reasoning behind this implementation.

Thank you for your time and assistance.