$Q\sharp$: Provably Optimal Distributional RL for LLM Post-Training
This addresses LLM alignment for improved reasoning, offering both performance gains and theoretical guarantees, though it builds incrementally on existing RL methods.
The paper tackles the problem of LLM post-training alignment by introducing Q♯, a value-based distributional RL algorithm that provably learns the optimal policy for KL-regularized RL. Empirically, it outperforms prior baselines in math reasoning benchmarks while maintaining smaller KL divergence to the reference policy.
Reinforcement learning (RL) post-training is crucial for LLM alignment and reasoning, but existing policy-based methods, such as PPO and DPO, can fall short of fixing shortcuts inherited from pre-training. In this work, we introduce $Q\sharp$, a value-based algorithm for KL-regularized RL that guides the reference policy using the optimal regularized $Q$ function. We propose to learn the optimal $Q$ function using distributional RL on an aggregated online dataset. Unlike prior value-based baselines that guide the model using unregularized $Q$-values, our method is theoretically principled and provably learns the optimal policy for the KL-regularized RL problem. Empirically, $Q\sharp$ outperforms prior baselines in math reasoning benchmarks while maintaining a smaller KL divergence to the reference policy. Theoretically, we establish a reduction from KL-regularized RL to no-regret online learning, providing the first bounds for deterministic MDPs under only realizability. Thanks to distributional RL, our bounds are also variance-dependent and converge faster when the reference policy has small variance. In sum, our results highlight $Q\sharp$ as an effective approach for post-training LLMs, offering both improved performance and theoretical guarantees. The code can be found at https://github.com/jinpz/q_sharp.