Brief · Thursday, Sep 3, 2026 · Day 34

Direct Preference Optimization: Your Language Model is Secretly a Reward Model

AI research · Rafailov et al. · 2023 · NeurIPS · 45 min · Dense
Summary

DPO shows that the reinforcement learning step of RLHF has a closed-form solution, so the policy can be trained directly on preference pairs with a simple classification-style loss. No separate reward model, no sampling loop, and results that match PPO-based training on the tasks studied.

Key ideas
  • The optimal RLHF policy can be written in terms of the reward and the reference model
  • Reparameterising the reward removes the need to learn it
  • The loss raises the likelihood of preferred answers relative to rejected ones, scaled by the reference
Why read it now

It replaced a fragile RL pipeline with one loss function and is now the default starting point for preference tuning.

Question to keep in mind

What does DPO give up compared with on-policy RL, and when would that matter?

124 online