Trends · 研究趋势
数据来自 Hugging Face 论文的 AI 提取关键词,按月统计研究方向的增长与热度。
Hao Liang, Mingrui Chen, Hengyi Feng +2 authors
DataFlex-RL evaluates reinforcement learning data policies and finds that uniform sampling matches or exceeds adaptive rollout selection, reweighting, and domain mixing across math, logic, and science benchmarks.
Julianna Piskorz, Antonin Berthon, Mihaela van der Schaar
In controlled distillation experiments, rollout policy has limited impact compared to token-level KL direction and learning rate, with on-policy benefits depending on objective and hyperparameters.
Shenzhi Wang, Le Yu, Chang Gao +15 authors
Token entropy patterns are crucial in RLVR, with high-entropy tokens significantly impacting reasoning performance and model optimization.
Fang Wu, Weihao Xuan, Heli Qi +4 authors
DeepSearch integrates Monte Carlo Tree Search into RLVR training to enhance exploration and credit assignment, achieving state-of-the-art performance with reduced computational cost.
Zixuan Huang, Yang Zhou, Kaixuan Wang +6 authors
Proposed multiple-choice planning and deferred trajectory verification reduce reasoning bias and improve verifiable autonomous driving decisions.
Yang Yue, Zhiqi Chen, Rui Lu +5 authors
Despite initial claims, reinforcement learning with verifiable rewards does not introduce fundamentally new reasoning abilities to LLMs, instead it enhances performance by biasing output distribution toward rewarded paths without expanding the reasoning boundary.
Xingxuan Li, Yao Xiao, Dianwen Ng +15 authors
The MiroMind-M1 series, built on Qwen-2.5, introduces fully open-source reasoning language models with state-of-the-art performance in mathematical reasoning through Context-Aware Multi-Stage Policy Optimization.
Guochao Jiang, Wenfeng Feng, Guofeng Quan +4 authors
A curriculum reinforcement learning framework dynamically adjusts training sample difficulty based on reward variance, improving LLM performance on mathematical reasoning tasks.
Yi Ding, Ruqi Zhang
On-policy distillation relies mainly on suppressing low-probability tokens rather than teacher guidance, motivating a supervision-free entropy-adaptive method that substantially improves reasoning performance.
Zheyuan Yang, Lyuhao Chen, Arman Cohan +1 authors
Two post-training strategies, distillation and RLVR, enable inference-time scaling in table reasoning tasks, resulting in a model (Table-R1-Zero) that matches GPT-4.1's performance using fewer parameters and shows strong generalization.
Fang Wu, Weihao Xuan, Ximing Lu +2 authors
Reinforcement Learning with Verifiable Rewards (RLVR) enhances precision but may limit exploration and discovery of new solutions, suggesting potential limits to its effectiveness in expanding reasoning capabilities.
Runzhe Zhan, Yafu Li, Zhi Wang +5 authors
ExGRPO, a framework that prioritizes valuable reasoning experiences, improves and stabilizes reinforcement learning from verifiable rewards for large language models.
Yifu Luo, Zeyu Chen, Haoyu Wang +4 authors
d-OPSD introduces a novel on-policy self-distillation framework for diffusion language models by adapting self-teacher construction and supervision mechanisms to match the non-autoregressive nature of diffusion models.
Jian Hu, Xibin Wu, Weixun Wang +3 authors
OpenRLHF is an open-source RLHF framework that enhances training efficiency and accessibility for researchers and practitioners.
Zhiqin Yang, Jingwen Fu, Yuhan Liu +16 authors
This work proposes a structured ladder for scaling large reasoning models beyond human supervision by tracing autonomous rewards and self-generated experience, while identifying risks and evaluation dimensions.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号