AMPO: Active Multi Preference Optimization for Self-play Preference Selection.

Taneesh Gupta, Rahul Madhavan, Xuchao Zhang, Chetan Bansal, Saravan Rajmohan

Browse the full ICML paper archive.