RL Post-Training

RL: Reinforcement Learning

Follow this topic Ask about this topic

Momentum

49 papers in the last four weeks, up 206% on the four weeks before. 0.5% of all new papers.

Jul 13Week of Sep 28

Latest papers 309

All topics

Beyond Compression: Diagnosing How Post-Training Changes Mathematical Reasoning
Sep 29, 2026Hongyang Li, Yiming Zhu, Xiao Li +3Language Model Distillation RL Post-Training
University of Luxembourg · Seafill Open-Source Community · Université Paris-Saclay
Where Does Staleness Accumulate? Pool Aware Effective Staleness Control for Asynchronous RL in LLM Post-Training
Sep 29, 2026Chenliang Li, Neiwen Ling, Zijun Wei +1Asynchronous RL RL for Language Model Reasoning
ByteDance · Texas A&M University · University of Virginia
ROSS: Relearning from Self-Generated Rollouts through Selective Supervision
Sep 28, 2026Zhiwei Zhang, Huayu Deng, Fei Zhao +4Supervised Fine-Tuning Self-Training for Language Models
AllSpark Team
PIVOT: Pivot-Aware On Policy Self Distillation for Multi-Turn VLM Agents
Sep 28, 2026Jiazhou Zhou, Hu Zhou, Yucheng Chen +3On-Policy Self-Distillation Reinforcement Learning with Verifiable Rewards
AI Thrust, The Hong Kong University of Science and Technology (Guangzhou) · International Digital Economy Academy (IDEA) · The Hong Kong Polytechnic University +2
Not All Rollouts Are Worth Learning: On Trajectory Valuation for Post-Training Reinforcement Learning
Sep 28, 2026Xuesong Jia, Ziao Yang, Zhanhe Huang +1RL Post-Training Data Valuation
Department of Computer Science, Brandeis University
Nereus: Adaptive Parallelism for LLM Post-Training
Sep 28, 2026Songlin Jiang, Tuo Shi, Sitong Zhang +3LLM Inference Scheduling RL Post-Training
Aalto University · Shenzhen University of Advanced Technology · Zhejiang University
The Low-Rank Structure of VLA Reinforcement Learning
Sep 28, 2026Minjae Oh, Yoonah Park, Jongwon Lim +1Vision-Language-Action Models RL Post-Training
Graduate School of Data Science, Seoul National University
FlexLoop: Depth-Elastic Looped Policies for Adaptive Test-Time Computation in Deep RL
Sep 28, 2026Xun Wang, Ruishuo Chen, Yu Chen +2Efficient Neural Network Inference Reinforcement Learning
Institute for Interdisciplinary Information Sciences, Tsinghua University
Coding Agent Memory Post-training: Unlocking the Memory Potential of Pre-trained File Operations for Long-Horizon Tasks via Reinforcement Learning
Sep 28, 2026Lirui Luo, Kelong Mao, Heming Xia +10LLM Agent Memory Persistent Memory for Language Models
Learning to Steer, Steering to See: Unveiling the Geometry of RLVR in Large Language Models via Trainable Vectors
Sep 28, 2026Yuchen Cai, Ding Cao, Qixiang Yin +10Reinforcement Learning with Verifiable Rewards RL Post-Training
USTC · Tencent Hunyuan · BUPT
Learning Perturbation Robust Policies for LLM Agents with Stable Optimization
Sep 28, 2026Pengxin Wang, Yuanzhe LI, Yuxin Ren +2RL Post-Training Robust RL
Department of Electrical and Computer Engineering University of Arizona
Selecting Diverse SFT Traces Improves Post-RL Generalization
Sep 27, 2026Dylan Zhang, Mingyuan Wu, Jinning LiReasoning Diversity RL for Language Model Reasoning
University of Illinois Urbana-Champaign, work done at Google · Google
Does Learning to Predict the World Help Agents Act? Auditing World-Model Post-Training
Sep 27, 2026Xinyu Che, Hang Yan, Yanchen Liu +5World Model Learning RL Post-Training
Xi’an Jiaotong University · University of Southern California · University of the Chinese Academy of Sciences +1
PoEM: Predicting RL Outcomes from Existing Policies
Sep 24, 2026Kimia Hamidieh, Giannis Daras, Antonio TorralbaRL for Generative Models RL Post-Training
MIT CSAIL
Rufus-Air: An Open LLM Post-Training Recipe
Sep 24, 2026Chia-Yuan Chang, Renyuan Cheng, Rui Feng +19LLM Fine-Tuning RL for Language Model Reasoning
Amazon
Transcript-Supervised Post-Training of Generative Speech Enhancement on Real Recordings via Reinforce Adjoint Matching
Sep 24, 2026Julius Richter, Christoph Boeddeker, Yoshiki Masuyama +4Speech Enhancement RL Post-Training
Mitsubishi Electric Research Laboratories (MERL), USA · Mitsubishi Electric Corporation, Japan
Not Every Token Is Worth Distilling: Selective Supervision for Direct-OPD
Sep 24, 2026Yibo Zhao, Zixuan Yang, Yunshi Lan +1Selective Knowledge Distillation On-Policy Distillation
School of Data Science and Engineering East China Normal University · Hugging Face
RL Starts before RL: On Policy Distillation for Better Reinforcement Learning
Sep 23, 2026Shuai Dong, Yongfu Zhu, Yuqi Xu +26Reinforcement Learning RL for Language Model Reasoning
Fudan University · SII · Peking University +2
PACT: From Credit Assignment to Critic Alignment
Sep 22, 2026Jiayan Fu, Hang Xu, Yong Zhang +4Reinforcement Learning Token-Level Credit Assignment
AllSpark Team
Reasoning-Preserving Fine-Tuning of Post-RL LLMs with Null-Basis LoRA
Sep 22, 2026Wenzhi Fang, Nicholas Tzou, Lazar Valkov +1LLM Fine-Tuning Low-Rank Adaptation
Samsung Research America, Mountain View, CA, USA · Purdue University, West Lafayette, IN, USA
Information-Time Proximal Policy Optimization
Sep 21, 2026Yongcheng Zeng, Xinyu Cui, Yan Song +9Reinforcement Learning RL for Language Model Reasoning
Institute of Automation, Chinese Academy of Sciences, China · School of Artificial Intelligence, University of Chinese Academy of Sciences, China · University College London +5
OPTED: On-Policy Fine-Tuning for End-to-End Driving using a Render-Free Teacher
Sep 17, 2026Damiano Da Col, Maximilian Igl, Peter Karkus +5End-to-End Autonomous Driving RL Post-Training
KE:SAI · ETH Zürich · NVIDIA Research +2
Compositional Reasoning in Language Models under Reinforcement Learning Post-Training
Sep 16, 2026Yu He, Yingxi Li, Yifei Wang +1Compositional Reasoning RL Post-Training
Stanford University · Amazon AGI Labs
CERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents
Sep 16, 2026Jiaxuan Jiang, Liyuan He, Zhixuan FangContinual Learning for LLM Agents Multi-Agent Systems
IIIS, Tsinghua University, Beijing, China · Shanghai Qi Zhi Institute, Shanghai, China · School of Artificial Intelligence, Shanghai Jiao Tong University, Shanghai, China
GrowMTP: Can RL Grow Its Own Draft Head?
Sep 15, 2026Minghua He, Lingzhe Zhang, Yuan Liu +2Speculative Decoding LLM Inference Acceleration
WeChat AI, Tencent · Peking University
Not All Prompts Are Equal: Exploration-Guided Prompt Scaffolding for Multimodal Reinforcement Post-Training
Sep 14, 2026Yuanhao Yue, Qianli Ma, Chengyu Wang +3Multimodal Large Language Models Curriculum RL
Alibaba Cloud Computing · Shanghai Jiao Tong University · Fudan University +1
Performance, Efficiency and Collapse -- Advantages and Challenges in Offline Post-training of Code LLMs
Sep 14, 2026Abhinav Anand, Sanjana Reddy Pachika, Shweta Verma +1RL for Code Generation Offline RL
TU Darmstadt · Hessian Center for Artificial Intelligence, Darmstadt, Germany · National Research Center for Applied Cybersecurity ATHENE
MInTRL: Off-policy Intervention can boost On-policy RL
Sep 14, 2026Mingyu Chen, Yefan Tao, Gerald Friedland +2Reinforcement Learning Reinforcement Learning with Verifiable Rewards
Amazon Web Services · Boston University
Expert-Space Exploration in MoE Reinforcement Learning
Sep 14, 2026Hongyi He, Zhenghao Lin, Xiao Liu +3Reinforcement Learning Expert Routing
Tsinghua University · Microsoft Research
Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-Training
Sep 9, 2026Junwon Ko, Dong-Jae Lee, Minchan Kwon +2Offline RL Preference Optimization
School of Electrical Engineering, KAIST Daejeon, Republic of Korea

Sep 29, 2026, cs.LGJ/K move · Enter open · S save

Beyond Compression: Diagnosing How Post-Training Changes Mathematical Reasoning

Hongyang Li, Yiming Zhu, Xiao Li, Caesar Wu+2

University of Luxembourg · Seafill Open-Source Community · Université Paris-Saclay

Language Model Distillation RL Post-Training LLM Post-Training LLM Mathematical Reasoning

Open paper AskPDF arXiv

Post-training is central to mathematical reasoning in modern large language models (LLMs), but endpoint pass@1 alone underidentifies what has changed. Gains may reflect newly reachable solutions, cheaper sampling of latent solutions, surface robustness, or memorisation. We compare three post-training paths under a common diagnostic readout: our sufficiently trained off-policy distillation trajectories, released Qwen3 off-policy-plus-on-policy distillation endpoints, and a released DeepSeek-Math endpoint trained with Group Relative Policy Optimisation (GRPO). Our probe uses cross-surface pass@K over verbatim prompts, paraphrases, numerical isomorphisms, and translations, plus consistency, distribution-shape, and verified supervised-fine-tuning (SFT) membership analyses. We find two regimes. On easier AMC problems, large-K ceilings are near saturation, so post-training mainly compresses sample cost. On harder AIME problems, post-training expands the large-K ceiling over the base model: sufficient off-policy distillation already raises this ceiling, Qwen3 released endpoints raise it further, and DeepSeek-Math GRPO does not dominate sufficient off-policy distillation at large K. English-dominant distillation improves non-English reasoning but preserves language-tier gaps. A controlled-overfit audit finds limited sensitivity in current SFT-membership probes. Compression is one regime of post-training, not a universal explanation.

Momentum

Related topics

Latest papers 309