Error when using pairwise: False in the judge config

[danielafrimi]

if setting the pairwise var to False, there is 1 game. when running the show_results.py script, it assumes 2 games are played.

in the get_battles_from_row function, this cause an index out of bound game = row["games"][1]

[CodingWithTim]

Hi @danielafrimi, is there a reason why you are setting pairwise to False? To run Arena-Hard as recommended, it should be pairwise comparison between baseline and each model answer. To avoid positional bias, we do 2 games each game with the position of the baseline and model answer swapped.