#ModelAccuracyPaper
1MetaAPO55.47link
2SELM55.21link
3Online DPO54.40link
4Chat OPM51.21link
5DPO48.62link
6SFT46.85link
7Base OPM45.35link
8Chat – (original)44.20link
9Chat OPM + XKD z (ours)42.90link
10Chat OPM + XKD h (ours)42.39link
11Chat OPM + UT42.20link
12Base – (original)40.06link
13Base OPM + XKD z (ours)39.74link
14Base OPM + XKD h (ours)39.33link
15Base OPM + UT39.11link
TruthfulQA truthfulqa Leaderboard