The paper studies reinforcement learning (RL) for post‑training language models on reasoning tasks, focusing on a hierarchical reward structure where each component becomes relevant only after the previous ones are resolved. It demonstrates that a Transformer‑based actor–critic algorithm that alternates between KL‑regularized policy sampling, critic fitting, and policy updates achieves minimax‑optimal rates in query budget and regularization strength, and is optimal for a fixed number of prompts. In contrast, sampling from a fixed reference distribution, as used in offline reward modeling, only yields a logarithmic regret decay, highlighting the advantage of on‑policy exploration in concentrating on high‑reward regions.
arXiv:2606. 00183v1 Announce Type: cross Abstract: Tree search is a central abstraction behind many language-agent reasoning and decision-making tasks: agents must explore actions, remember failures, and backtrack toward promising alternatives.
arXiv:2607. 16205v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards has emerged as a standard approach for enhancing reasoning in large language models, which typically optimizes the policy by contrasting multiple self generated rollouts.
arXiv:2609.37119v1 Announce Type: cross Abstract: Recent approaches to reinforcement learning (RL) post-training for large language models increasingly remove the critic to reduce training instabilit...
The paper introduces POISE, a reinforcement learning algorithm that uses a model’s internal states as a value estimator to reduce variance in reinforcement learning with verifiable rewards (RLVR). By employing a lightweight probe that reads internal signals during the forward pass, POISE predicts baselines online and uses a cross‑rollout construction to keep gradients unbiased. Experiments on Qwen3‑4B and OLMo3‑7B‑Instruct‑DPO across six domains show POISE outperforms existing RLVR baselines, offering more stable training and a value model that generalizes across tasks and scales with the policy.
arXiv:2608. 09805v1 Announce Type: cross Abstract: Exploration has been a focus of reinforcement learning research for a long time.
Achieving strong optimization generalization across diverse optimization problems while requiring limited training resources remains a challenging problem for optimization-oriented large language models (LLMs). Existing approaches typically rely on large-scale supervised datasets, costly reasoning annotations, and expensive intermediate step verification, resulting in substantial training overhead.
arXiv:2607. 04713v1 Announce Type: cross Abstract: Reinforcement learning holds significant potential for training large language models (LLMs) to handle multi-turn interactive tasks.
The paper "Demystifying Reinforcement Learning Post-Training of Language Models" investigates how reinforcement learning (RL) post‑training enhances large language models (LLMs) for tasks such as reasoning, math, and coding. By isolating RL components in a controlled setting, the authors analyze how the base model’s prior distribution, reward granularity, prompt diversity, and model scale influence outcomes, using policy entropy to compare pre‑training, supervised fine‑tuning (SFT), and RL stages. The study clarifies the role of spurious rewards, the importance of the base model’s probability mass on desired behaviors, and how these factors interact to determine post‑training success, offering a practical primer for NLP researchers. "whyItMatters":"The work provides a clearer understanding of RL post‑training mechanics, helping researchers and practitioners effectively apply RL to improve LLM capabilities."
The paper explores how dense, turn-level reward structures can improve reinforcement learning for large language model agents in multi-turn tasks. It introduces three reward granularity types—terminal, delayed, and per-turn—and adapts Group Relative Policy Optimization and Proximal Policy Optimization to each. Experiments on search and game agents show that per-turn rewards consistently yield better training dynamics, faster convergence, and higher answer correctness compared to sparse terminal or delayed rewards.
Rationale-Guided Policy Optimization (RGPO) is a reinforcement‑learning framework that adaptively uses ground‑truth rationale information to scaffold a language model’s reasoning process. Instead of treating reference solutions as fixed imitation targets, RGPO temporarily incorporates rationales to help the model generate better responses, then reverts to unguided learning with higher‑reward, model‑generated solutions. Experiments in both language‑only and vision‑language tasks show that RGPO consistently outperforms RLVR baselines, with ablation studies confirming that adaptive rationale guidance is a key factor in its success.
arXiv:2601. 15158v4 Announce Type: replace-cross Abstract: Transformers trained via Reinforcement Learning (RL) with outcome-based supervision can spontaneously develop the ability to generate intermediate reasoning steps (Chain-of-Thought).
arXiv:2607. 06987v1 Announce Type: new Abstract: Reinforcement learning (RL) has become the standard paradigm for enhancing the complex reasoning capabilities of large language models (LLMs).