1
The Bidirectional Process Reward Model
提出双向过程奖励模型，突破传统单向奖励局限，提升语言模型推理与对齐性能。
arXiv:2508.01682v3 Announce Type: replace Abstract: Process Reward Models (PRMs), which assign fine-grained scores to intermediate reasoning steps wit…
2
Steerable Cultural Preference Optimization of Reward Models
提出可引导文化偏好优化方法，让奖励模型对齐多元文化价值观，突破单一偏好局限。
arXiv:2606.18606v1 Announce Type: new Abstract: It is essential for large language model (LLM) technology to serve many different cultural sub-communi…
3
The Coin Flip Judge? Reliability and Bias in LLM-as-a-Judge Evaluation
LLM评判模型输出的可靠性如同抛硬币？这篇研究系统揭示评估过程中的偏见与随机性问题。
arXiv:2606.13685v1 Announce Type: cross Abstract: LLM-as-a-Judge is now widely used to rank model outputs, train reward models, and populate public le…
4
Improving Heart-Focused Medical Question Answering in LLMs via Variance-Aware Rubric Rewards with GRPO
用方差感知评分奖励与GRPO方法提升大模型心脏医学问答准确性，为医疗AI提供新思路。
arXiv:2606.05174v1 Announce Type: cross Abstract: Large Language Models (LLMs) have shown strong promise in healthcare applications. Yet deploying gen…
5
Test-time reward-guided alignment of language models by importance sampling on pre-logit space
无需重新训练，在推理阶段通过pre-logit空间重要性采样实现奖励对齐，高效且不影响模型原有能力。
arXiv:2510.26219v3 Announce Type: replace-cross Abstract: Test-time alignment of large language models (LLMs) attracts attention because fine-tuning o…
6
Configurable Reward Model for Balanced Safety Alignment
可配置奖励模型实现安全性与实用性的动态平衡，为大模型对齐难题提供新解法
arXiv:2605.30487v1 Announce Type: new Abstract: Aligning large language models (LLMs) to heterogeneous and rapidly evolving safety requirements remain…
7
RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains
提出RUBRIC-ARROW方法，通过交替点对点标准奖励建模优化LLM在非可验证领域的后训练性能
arXiv:2605.29156v1 Announce Type: new Abstract: Pointwise reward modeling offers critical signals for LLM post-training, yet struggles with absolute s…
8
Verifiable Process Rewards for Agentic Reasoning
提出可验证过程奖励机制，让智能体推理更可信可解释，强化学习新思路。
arXiv:2605.10325v2 Announce Type: replace Abstract: Reinforcement learning from verifiable rewards (RLVR) has improved the reasoning abilities of larg…
9
KARMA: Karma-Aligned Reward Model Adaptation
论文提出KARMA方法，创新地将业力概念融入奖励模型对齐，为AI伦理对齐提供新思路。
arXiv:2605.26738v1 Announce Type: new Abstract: Human communication depends on implicit social signals where effectiveness is shaped by tone, context,…
10
Understanding and Mitigating Premature Confidence for Better LLM Reasoning
最新研究揭示LLM长思维链中“过早自信”导致的逻辑缺口，并提出基于过程奖励模型的缓解策略，提升推理质量。
arXiv:2605.24396v1 Announce Type: new Abstract: Long chains of thought (CoT) from current language models frequently contain logical gaps and unjustif…
11
VRPRM: Process Reward Modeling via Visual Reasoning
通过视觉推理提升过程奖励建模精度，为复杂任务训练提供新思路。
arXiv:2508.03556v3 Announce Type: replace Abstract: Process Reward Model (PRM) is widely used in the post-training of Large Language Model (LLM) becau…
12
Leveraging Verifier-Based Reinforcement Learning in Image Editing
将RLHF引入图像编辑的新范式，提出基于验证器的强化学习解决奖励模型缺失瓶颈。
arXiv:2604.27505v2 Announce Type: replace Abstract: While Reinforcement Learning from Human Feedback (RLHF) has become a pivotal paradigm for text-to-…
13
rePIRL: Learn PRM with Inverse RL for LLM Reasoning
用逆强化学习从推理轨迹中自动学习过程奖励模型，有效提升大语言模型的复杂推理能力。
arXiv:2602.07832v2 Announce Type: replace Abstract: Process rewards have been widely used in deep reinforcement learning to improve training efficienc…
14
Eliminating Inductive Bias in Reward Models with Information-Theoretic Guidance
ICLR 2026 顶会论文：用信息论指导消除奖励模型中的归纳偏置，为强化学习对齐提供更客观的评估基础
arXiv:2512.23461v2 Announce Type: replace Abstract: Reward models (RMs) are essential in reinforcement learning from human feedback (RLHF) to align la…
15
CodeScaler: Scaling Code LLM Training and Test-Time Inference via Reward Models
用奖励模型突破测试用例限制，实现代码大模型训练与推理阶段的可扩展强化学习。
arXiv:2602.17684v2 Announce Type: replace Abstract: Reinforcement Learning from Verifiable Rewards (RLVR) has driven recent progress in code large lan…