Rethinking the Trust Region in LLM Reinforcement Learning
This addresses a core problem in RL-based LLM fine-tuning for researchers and practitioners, offering a more robust foundation, though it is incremental as it builds on PPO.
The paper tackled the inefficiency and instability in fine-tuning Large Language Models (LLMs) with Proximal Policy Optimization (PPO) by proposing Divergence Proximal Policy Optimization (DPPO), which replaces heuristic clipping with a principled divergence constraint and achieves superior training stability and efficiency.
Reinforcement learning (RL) has become a cornerstone for fine-tuning Large Language Models (LLMs), with Proximal Policy Optimization (PPO) serving as the de facto standard algorithm. Despite its ubiquity, we argue that the core ratio clipping mechanism in PPO is structurally ill-suited for the large vocabularies inherent to LLMs. PPO constrains policy updates based on the probability ratio of sampled tokens, which serves as a noisy single-sample Monte Carlo estimate of the true policy divergence. This creates a sub-optimal learning dynamic: updates to low-probability tokens are aggressively over-penalized, while potentially catastrophic shifts in high-probability tokens are under-constrained, leading to training inefficiency and instability. To address this, we propose Divergence Proximal Policy Optimization (DPPO), which substitutes heuristic clipping with a more principled constraint based on a direct estimate of policy divergence (e.g., Total Variation or KL). To avoid huge memory footprint, we introduce the efficient Binary and Top-K approximations to capture the essential divergence with negligible overhead. Extensive empirical evaluations demonstrate that DPPO achieves superior training stability and efficiency compared to existing methods, offering a more robust foundation for RL-based LLM fine-tuning.