Yijia Fan Nanyang Technological University, Microsoft Research Asia and SUN YAT-SEN UNIVERSITY, Jing Yang, Mingyu Liu, Kaitong Cai, Jian Wang Snap Inc., Keze Wang SUN YAT-SEN UNIVERSITY, Jusheng Zhang
Diffusion Large Language Models (dLLMs) have emerged as a promising non-autoregressive paradigm for text generation, offering parallel decoding and bidirectional context modeling. However, aligning dLLMs with reinforcement learning (RL) remains a significant challenge, as the marginal likelihood of sequences in masked diffusion is typically intractable, rendering standard policy gradient methods unstable or computationally prohibitive. In this work, we propose Diffusion-Gibbs Alignment (DGA), a novel variational framework that reformulates RL for dLLMs as a distribution matching problem. DGA bypasses the explicit computation of log-probabilities by leveraging a learned energy function to model the relative quality of samples. The optimization is decoupled into two stable steps: (1) contrastive energy ranking to capture global reward structures, and (2) weighted diffusion alignment to update the policy via importance sampling. Empirically, DGA establishes a new state-of-the-art across logical reasoning (Sudoku, Countdown), mathematical reasoning (GSM8K, Math500), and code generation (HumanEval, MBPP) benchmarks. DGA offers a novel variational perspective for dLLM alignment, achieving better performance while simultaneously enhancing training speed and memory efficiency.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Kaitong Cai, Jusheng Zhang, Keze Wang SUN YAT-SEN UNIVERSITY
Hybrid offline–online reinforcement learning (O2O RL) promises both sample efficiency and robust exploration, but suffers from instability due to distribution shift between offline and online data. We introduce RLPD-GX, a framework that decouples policy optimization from safety enforcement: a reward-seeking learner explores freely, while a projection-based guardian guarantees rule-consistent execution and safe value backups. This design preserves the exploratory value of online interactions without collapsing to conservative policies. To further stabilize training, we propose dynamic curricula that gradually extend temporal horizons and anneal offline–online data mixing. We prove convergence via a contraction property of the guarded Bellman operator, and empirically show state-of-the-art performance on Atari-100k, achieving a normalized mean score of 3.02 (+45% over prior hybrid methods) with stronger safety and stability. Beyond Atari, ablations demonstrate consistent gains across safety-critical and long-horizon tasks, underscoring the generality of our design. Extensive and comprehensive results highlight decoupled safety enforcement as a simple yet principled route to robust O2O RL, suggesting a broader paradigm for reconciling exploration and safety in reinforcement learning.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表