Return to Microsoft Research Lab – Redmond
AI Interaction and Learning
Articles
Chargement…
Articles
Direct Nash Optimization: Teaching language models to self-improve with general preferences
This talk discusses teaching language models to self-improve using a preference oracle like GPT-4, framing it as a two-player game to find an optimal policy at a Nash equilibrium, and achieving state-of-the-art win rates against GPT-4 Turbo on benchmarks such…