Exploratory Preference Optimization: Harnessing Implicit Q*-Approximation for Sample-Efficient RLHF

Documenti analoghi

• Direct Nash Optimization: Teaching Language Models to Self-Improve with General Preferences
  di: Rosset, Corby, et al.
  Pubblicazione: (2024)
• Orca-Math: Unlocking the potential of SLMs in Grade School Math
  di: Mitra, Arindam, et al.
  Pubblicazione: (2024)
• Correcting the Mythos of KL-Regularization: Direct Alignment without Overoptimization via Chi-Squared Preference Optimization
  di: Huang, Audrey, et al.
  Pubblicazione: (2024)
• Active Preference Optimization for Sample Efficient RLHF
  di: Das, Nirjhar, et al.
  Pubblicazione: (2024)
• Trajectory Bellman Residual Minimization: A Simple Value-Based Method for LLM Reasoning
  di: Yuan, Yurun, et al.
  Pubblicazione: (2025)