中文
相关论文

相关论文: On the Hidden Objective Biases of Group-based Rein…

200 篇论文

Despite their sophisticated general-purpose capabilities, Large Language Models (LLMs) often fail to align with diverse individual preferences because standard post-training methods, like Reinforcement Learning with Human Feedback (RLHF),…

Post-training plays a crucial role in refining and aligning large language models to meet specific tasks and human preferences. While recent advancements in post-training techniques, such as Group Relative Policy Optimization (GRPO),…

人工智能 · 计算机科学 2025-10-28 Kaichen Zhang , Yuzhong Hong , Junwei Bao , Hongfei Jiang , Yang Song , Dingqian Hong , Hui Xiong

Group Relative Policy Optimization (GRPO) has shown promise in discrete action spaces by eliminating value function dependencies through group-based advantage estimation. However, its application to continuous control remains unexplored,…

机器人学 · 计算机科学 2025-07-29 Rajat Khanda , Mohammad Baqar , Sambuddha Chakrabarti , Satyasaran Changdar

We revisit Group Relative Policy Optimization (GRPO) in both on-policy and off-policy optimization regimes. Our motivation comes from recent work on off-policy Proximal Policy Optimization (PPO), which improves training stability, sampling…

The Group Relative Policy Optimization (GRPO), a reinforcement learning method used to fine-tune large language models (LLMs), has proved its effectiveness in practical applications such as DeepSeek-R1. It raises a question whether GRPO can…

机器学习 · 计算机科学 2025-11-20 Yanchen Xu , Ziheng Jiao , Hongyuan Zhang , Xuelong Li

Reinforcement learning with verifiable rewards (RLVR) has become a practical route to improve large language model reasoning, and Group Relative Policy Optimization (GRPO) is a widely used optimizer in this setting. However, RLVR training…

机器学习 · 计算机科学 2026-05-14 Tue Le , Linh Ngo Van , Trung Le

As language models become increasingly capable, users expect them to provide not only accurate responses but also behaviors aligned with diverse human preferences across a variety of scenarios. To achieve this, Reinforcement learning (RL)…

Standard reinforcement learning from human feedback (RLHF) trains a reward model on pairwise preference data and then uses it for policy optimization. However, while reward models are optimized to capture relative preferences, existing…

机器学习 · 计算机科学 2026-02-05 Kyuseong Choi , Dwaipayan Saha , Woojeong Kim , Anish Agarwal , Raaz Dwivedi

Group Relative Policy Optimisation (GRPO) enhances large language models by estimating advantages across a group of sampled trajectories. However, mapping these trajectory-level advantages to policy updates requires aggregating token-level…

Hybrid Group Relative Policy Optimization (Hybrid GRPO) is a reinforcement learning framework that extends Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO) by incorporating empirical multi-sample action…

机器学习 · 计算机科学 2025-02-05 Soham Sane

Generative speech enhancement offers a promising alternative to traditional discriminative methods by modeling the distribution of clean speech conditioned on noisy inputs. Post-training alignment via reinforcement learning (RL) effectively…

音频与语音处理 · 电气工程与系统科学 2026-01-26 Haoxu Wang , Biao Tian , Yiheng Jiang , Zexu Pan , Shengkui Zhao , Bin Ma , Daren Chen , Xiangang Li

Group Relative Policy Optimization(GRPO) has become a cornerstone of modern reinforcement learning alignment, prized for its efficacy in foregoing an explicit value-critic by leveraging reward normalization across sampled trajectory…

计算与语言 · 计算机科学 2026-05-29 Redacted by arXiv

Group-Relative Policy Optimization (GRPO) has emerged as the standard for training reasoning capabilities in large language models through reinforcement learning. By estimating advantages using group-mean rewards rather than a learned…

人工智能 · 计算机科学 2026-03-06 Anisha Garg , Claire Zhang , Nishit Neema , David Bick , Ganesh Venkatesh , Joel Hestness

Group Relative Policy Optimization (GRPO) has emerged as a popular algorithm for reinforcement learning with large language models (LLMs). However, upon analyzing its clipping mechanism, we argue that it is suboptimal in certain scenarios.…

机器学习 · 计算机科学 2026-01-08 Chi Liu , Xin Chen

Group Relative Policy Optimization (GRPO), which is widely adopted by R1-like reasoning models, has advanced mathematical reasoning. Nevertheless, GRPO faces challenges in reward sparsity, verbosity, and inadequate focus on problem…

计算与语言 · 计算机科学 2025-09-23 Jixiao Zhang , Chunsheng Zuo

Group Relative Policy Optimization (GRPO) was introduced and used recently for promoting reasoning in LLMs under verifiable (binary) rewards. We show that the mean + variance calibration of these rewards induces a weighted contrastive loss…

机器学习 · 计算机科学 2025-10-22 Youssef Mroueh

Recent advances in large language models (LLMs) have broadened their applicability across diverse tasks, yet specialized domains still require targeted post training. Among existing methods, Group Relative Policy Optimization (GRPO) stands…

机器学习 · 计算机科学 2025-08-08 Ziyin Gu , Jingyao Wang , Ran Zuo , Chuxiong Sun , Zeen Song , Changwen Zheng , Wenwen Qiang

Reinforcement Learning with Human Feedback (RLHF) has been the dominant approach for improving the reasoning capabilities of Large Language Models (LLMs). Recently, Reinforcement Learning with Verifiable Rewards (RLVR) has simplified this…

计算与语言 · 计算机科学 2025-10-10 Yining Wang , Jinman Zhao , Chuangxin Zhao , Shuhao Guan , Gerald Penn , Shinan Liu

Aligning large language models (LLMs) with human values and safety constraints is challenging, especially when objectives like helpfulness, truthfulness, and avoidance of harm conflict. Reinforcement Learning from Human Feedback (RLHF) has…

计算与语言 · 计算机科学 2025-03-31 Xuying Li , Zhuo Li , Yuji Kosuga , Victor Bian

Since DeepSeek-R1 popularized, Group Relative Policy Optimization (GRPO) has become the core part of training Reasoning LLMs. However, we find some deficiency that influences RL stability and inference efficiency, like zero-variance in…

计算与语言 · 计算机科学 2025-09-30 Chen Li , Nazhou Liu , Kai Yang
‹ 上一页 1 2 3 10 下一页 ›