RL for Tool Use

RL: Reinforcement Learning

Follow this topic Ask about this topic

Momentum

17 papers in the last four weeks, up 42% on the four weeks before. 0.2% of all new papers.

Jul 13Week of Sep 28

Latest papers 127

All topics

Self-Evolve With a Reference:Anchored Training of Tool-Integrated Agents
Oct 7, 2026Wenjie Liao, Liangjie Zhao, Zehong CaoLLM Agent Self-Improvement Self-Evolving Agents
Waseda University · Adelaide University
FC-SWE: Failure-Conditioned RL for Long-Horizon Software Engineering Agents
Oct 6, 2026Jia Liufu, Bin Hu, Linglin Jing +5RL for Code Generation Software Engineering Agents
NVIDIA · Tsinghua University
CIPO: Counterfactual Imagination Policy Optimization for Adaptive Tool Granularity Selection
Oct 4, 2026Yu Li, Yunlu Wan, Zijian Zhu +4LLM Agent Skill Learning RL for Tool Use
School of Computer Science and Engineering, Southeast University, Nanjing, China · KTH Royal Institute of Technology · Huawei Noah’s Ark Lab, China
HybridCUA: Learning to Orchestrate GUI and CLI for Computer-Use Agents
Sep 29, 2026Tongbo Chen, Junbo Niu, Zhengxi Lu +8Computer-Use Agents GUI Agents
Zhejiang University · Peking University · Tsinghua University
RLTL;DR: Self-improvement by Internalizing Self-generated Feedback
Sep 29, 2026Michael Kirchhof, Eleonora Gualdoni, Andrew Szot +6RL for Code Generation Reinforcement Learning with Verifiable Rewards
Apple
Neuro-Symbolic Computer Use: Learning Reusable Policies for Reliable and Efficient Execution
Sep 29, 2026Hyewon Suh, Thanh Minh Nguyen, Chih-Lun Lee +5Computer-Use Agents Runtime Enforcement for AI Agents
Georgia Institute of Technology · Simular · Stanford University
WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents
Sep 29, 2026Bo Mao, Hang He, Linting Wang +17LLM Agent Self-Improvement LLM Post-Training
East China Normal University · Shanghai Innovation Institute · Fudan University +2
MLToolBench: Learning Tool-Augmented Agents for Machine Learning Development
Sep 29, 2026Xin Yu, Lizhu Zhang, Jiamu Bai +7LLM Agent Skill Learning AI Agent Benchmarks
ParaAgent: Reinforcing Parallel Acting in Open-World Tool Environments
Sep 27, 2026Shengbin Yue, Hongru Wang, Siyuan Wang +3LLM Agent Training Tool Use
Fudan University · University of Edinburgh · Chinese University of Hong Kong +3
SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL
Sep 24, 2026Yan Zhan, Shaobo Liu, Qiunan Liu +7Group Relative Policy Optimization Credit Assignment in RL
Peking University · Shenzhen University · Tencent PCG QQ Team
When Does Action Credit Need Updating?
Sep 24, 2026Hongye Yang, Boxiao HuangCredit Assignment in RL Credit Assignment
College of Computing Georgia Institute of Technology
Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use
Sep 21, 2026Zixiang Chen, Wenting Zhao, Zhepeng Cen +9Reinforcement Learning Credit Assignment in RL
Salesforce AI Research
UnifiedPlayers: Enhance Tool-Integrated Reasoning in Agentic Reinforcement Learning
Sep 17, 2026Wenjie Liao, Liangjie Zhao, Zehong CaoRL for Language Model Reasoning Reinforcement Learning with Verifiable Rewards
Waseda University · Adelaide University
MATCH: Model-Aware Tool Learning with Curriculum Scheduling and Hierarchically Gated Rewards
Sep 17, 2026Shihao Liu, Hao Yin, Lijun Liu +3Reward Modeling Curriculum RL
School of Cyber Security, University of the Chinese Academy of Sciences · Honor Device Co., Ltd · University of the Chinese Academy of Sciences
SFT or RL for Tool-Calling Agents? A Controlled Study Across Data, Method, and Scale
Sep 15, 2026Md Tahmid Rahman Laskar, Xue-Yong Fu, Shashi Bhushan TNSupervised Fine-Tuning LLM Tool Use
Dialpad Inc.
EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents
Sep 15, 2026Sehee Kim, Yumin Choi, Minki Kang +1LLM Agent Self-Improvement Prompt Optimization
KAIST
Spurious Tool Use: When RL Agents Learn the Wrong Reason to Act
Sep 14, 2026Yiwei Yang, Haoxiang Zhang, Bingbing Wen +6LLM Agent Reliability RL for Tool Use
University of Washington · University of California San Diego · Stanford University
T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks
Sep 12, 2026Junyao Yang, Yucheng Shi, Zhongzhi Li +4Terminal Agents Long-Horizon Agent Evaluation
Tencent Hy Foundation Model · National University of Singapore · University of Georgia +2
Why Sample What You Can Enumerate? Exact Policy Optimization for Genomic Tool Selection
Sep 9, 2026Haoyue Liu, Xiaoyu Ma, Ye Chen +2RL for Language Model Reasoning Group Relative Policy Optimization
School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen 518172, China · Shenzhen Future Network of Intelligence Institute (FNii-Shenzhen) · XJTU-POLIMI Joint School, Xi’an Jiaotong University, Xi’an 710049, China
Learning to Zoom Efficiently with a Contrastive Curriculum
Sep 2, 2026Falko Helm, Iryna GurevychTool Use in VLMs Contrastive RL
Ubiquitous Knowledge Processing Lab (UKP Lab) Department of Computer Science and Hessian Center for AI (hessian.AI) Technical University of Darmstadt
ARISE-RL: Agentic Rubric-Grounded Iterative Self-Evolution with Reinforcement Learning
Sep 1, 2026Fanrui Zhang, Ruixue Ding, Qiang Zhang +13AI Agent Benchmarks Rubric-Based RL
Alibaba ATH Token Foundry · Hema, Alibaba Group
CoBRA: Learning Tool-Use Boundaries via Counterfactual Margins
Sep 1, 2026Wenhao Zou, Xianglong Liu, Wendong Bi +3Tool Use RL for Tool Use
WeChat, Tencent Inc, Beijing, China · University of Chinese Academy of Sciences, Beijing, China · Wuhan University, Wuhan, China
One Policy, Any Budget: Internalizing Budget-Aware Search via Reinforcement Learning
Sep 1, 2026Xiaowei Sun, Jin Li, Yili Hong +2Cost-Aware Inference Constrained RL
College of Computer Science and Artiﬁcial Intelligence, Fudan University · College of Software Engineering, Southeast University · School of Artiﬁcial Intelligence, Shanghai Jiao Tong University
One Policy Is Enough: Single-Agent Reinforcement Learning Outperforms Tree Search for Chemistry Tool Learning
Aug 31, 2026Armin Dariani, Sifan Wu, Bang Liu +1LLM Tool Use Tool Use
1DIRO, Université de Montréal · 2Mila - Quebec Artificial Intelligence Institute · 3Innovation Campus Delaware, Air Liquide
Beyond Task Completion: Training Capable and Safe Computer-Use Agents
Aug 27, 2026Zeyu Kang, Zhenyun Yin, Yang Zhang +5Computer-Use Agents AI Agent Safety
Shanghai Artificial Intelligence Laboratory · Harbin Institute of Technology · Fudan University +2
HiDiffTIR: Hierarchical Difficulty-Aware Policy Optimization for Multi-Turn Tool-Integrated Reasoning
Aug 22, 2026Yucan Guo, Xiaohan Wang, Miao Su +8Credit Assignment RL for LLM Agents
State Key Laboratory of AI Safety · Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy of Sciences +1
SPADE: Self-Play in Adaptive Synthetic Executable Environments
Aug 19, 2026Bo Liu, Simon Yu, Yiding Jiang +15Self-Play RL LLM Agent Self-Improvement
University of Washington · Stanford University · Northeastern University +6
RTPO: Reverse-Turn Policy Optimization for Stabilizing Agentic RL Training
Aug 19, 2026Yugu Li, Zehong Cao, Jianglin Qiao +1Agentic RL Reinforcement Learning
School of CSIT · ACFR · School of EECMS
Teach the Magnitude, Not the Direction: Verifier-Bounded Credit Assignment for Multi-Turn Multi-step LLM Agents
Aug 13, 2026Zechuan Wang, Siyuan Lu, Hongxuan Zhang +3Token-Level Credit Assignment Step-Level Credit Assignment in RL
Zhejiang University · AWorld Team, Inclusion AI · Shanghai Innovation Institute +2
Retry, Switch, or Abstain? Learning Strategy-Aware Tool-Use Policies via Controlled Error Injection
Aug 12, 2026Chaoran Chen, Vy Nguyen, Ziji Zhang +7Curriculum RL AI Agent Benchmarks
Amazon

Oct 7, 2026, cs.AIJ/K move · Enter open · S save

Self-Evolve With a Reference:Anchored Training of Tool-Integrated Agents

Wenjie Liao, Liangjie Zhao, Zehong Cao

Waseda University · Adelaide University

LLM Agent Self-Improvement Self-Evolving Agents Curriculum RL RL for Tool Use

Open paper AskPDF arXiv

Self-evolving tool-integrated agents learn from tasks and feedback generated within their own training loop. A Curriculum Agent generates tasks, while an Executor Agent learns from self-consistency signals through reinforcement learning. However, relying solely on the current Executor for feedback has two limitations: group-relative advantages vanish under full consensus, while uncertainty-based curriculum rewards favor disagreement without showing whether the generated tasks support further learning. These limitations motivate an additional reference beyond the current Executor. We propose \textit{AnchorLoop}, which introduces a frozen copy of the previous iteration's Executor as a historical reference and reuses it on both sides of the training loop. For the Executor, the anchor provides a cross-reference advantage that evaluates current outputs against both current and historical majority answers. For the Curriculum, it provides an agreement-based reference based on differences in sampled majority agreement. Since the Executor and anchor have identical parameters during Curriculum training, this comparison serves as a proxy for task selection rather than evidence of inter-version improvement or correctness. Across 13 reasoning benchmarks, AnchorLoop improves over Agent0 by 2.5% on mathematical reasoning and 2.8% on general reasoning tasks. It also maintains higher effective-advantage variance and continues improving in later iterations as the unanchored baseline shows diminishing gains. These results demonstrate the benefit of introducing a lightweight historical reference into self-evolving tool-integrated agents without external task or answer supervision.

Momentum

Related topics

Latest papers 127