Generalization in RL

RL: Reinforcement Learning

The field asks whether an agent can act effectively when goals, layouts, visual inputs, dynamics, or collaborators differ from training, a requirement for reliable deployment. Common approaches include domain randomization and exploration for broader state coverage, invariant or task-conditioned representations, meta-learning, curricula, reusable skills, and Bayesian or model-based transfer. Researchers evaluate these methods in settings such as Procgen, MuJoCo, MetaWorld, DeepMind Control, robotics, and language-based agents, using held-out tasks, environments, and visual or dynamics shifts. A central tension is that diversity and regularization can improve transfer but hinder learning or retention, while benchmark design and metrics can make apparent generalization gains hard to interpret.

Latest papers

89 papers16 papers in the last four weeks, up 60% on the four weeks before. 0.1% of all new papers.

Harness Scaling for Generalizable Agentic Reinforcement Learning
Oct 21, 2026Fei Bai, Huatong Song, Chuan Hao +5Agentic RL LLM Agent Harnesses
Gaoling School of Artificial Intelligence, Renmin University of China · IQuest Research · Beihang University
Reframing Object-Centric Reinforcement Learning: From Absolute to Relative Representations
Oct 20, 2026Cedric Derstroff, Jannis Blüml, Janosch Moos +4Reinforcement Learning Representation Learning for RL
The Hessian Center for Artificial Intelligence (hessian.AI) · Technische Universität Darmstadt · Centre for Cognitive Science +1
Co-Evolving Policy and Reward for Code Generation: A Mechanistic Study of Single-Adapter Drift
Oct 19, 2026Amarsaikhan Batjargal, Noriko TomuroRL for Code Generation Reinforcement Learning
Jarvis College of Computing and Digital Media, DePaul University, Chicago, IL, USA
Robust Reinforcement Learning for Dynamic Treatment Regimes under Model and Measurement Uncertainty
Oct 15, 2026Chengze Xie, Xuan Wang, Ying Ding +2Robust RL Sequential Decision Making
Department of Industrial and Systems Engineering, Texas A&M University. · Department of Mechanical Engineering, The University of Texas at Austin. · School of Information, The University of Texas at Austin.
What Happens to the Goal? Distinct Failures of Policy Alignment in Goal Misgeneralization
Oct 14, 2026Julia Barth, Armin Sommer, Kunvar Thaman +1AI Alignment Generalization in RL
Pantheon Industries · ETH Zurich
Imagining Goals with Geometric Horizon Models for Policy Generalization
Oct 13, 2026Daniel Lawson, Dhruv Sreenivas, Adriana Hugessen +2Robot Policy Generalization Generalization in RL
Mila · Université de Montréal · Google DeepMind
An Informational Curse of Horizon in Goal-Conditioned Policy Learning
Oct 7, 2026John L. Zhou, Yuxuan Dong, Jonathan C. KaoGoal-Conditioned RL Generalization in RL
University of California, Los Angeles
Considering Context: When World Models Need Context Encoders
Oct 5, 2026Oleg Smirnov, Sofiane Ennadir, John Pertoft +2Reinforcement Learning World Model Learning
King AI Labs, Microsoft Gaming
How RL Reshapes LLM Reasoning: Transferability, Coverage, and Scaling Laws
Oct 3, 2026Ziheng Cheng, Yixiao Huang, Hanlin Zhu +1RL for Language Model Reasoning Reinforcement Learning
University of California, Berkeley
Mastering Atari 2600 Games with Discovered Options
Oct 2, 2026Erik M. Lintunen, Marlos C. MachadoRL for Game AI Hierarchical RL
University of Alberta · Alberta Machine Intelligence Institute (Amii) · Canada CIFAR AI Chair
VIGOR: Zero-Shot Visual Generalization via Latent-Space Consistency in Model-Based Reinforcement Learning
Oct 2, 2026Mingyu Park, Samyeul Noh, Hyun Myung +1Latent-Space Regularization Generalization in RL
Robotics Program KAIST, ETRI · Electrical Engineering KAIST
Reward as Observation: Learning Reward-Based Policies for Rapid Adaptation
Sep 30, 2026Morgan Byrd, Jacob Blevins, Maks Sorokin +2Generalization in RL Robot Policy Adaptation
Georgia Institute of Technology, Atlanta, GA, 30308, USA · Georgia Tech Research Institute, Atlanta, GA, 30308, USA
Learning Transferable Skills using Goal-Conditioned Bisimulation
Sep 30, 2026Mohammad Amin Abbasfar, Farbod Azimmohseni, Mohammad Hossein RohbanAgent Skill Learning Trajectory Representation Learning
Department of Computer Engineering Sharif University of Technology
EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making
Sep 29, 2026Yuhan Guo, Jinming Liu, Liang Xu +8LLM Agent Training Generalization in RL
Shanghai Jiaotong University · Eastern Institute of Technology, Ningbo · Zhongguancun Academy, Beijing, China +1
On-Policy Parameter Update Direction Underlies Generalization in LLM Post-Training
Sep 29, 2026Shufan Shen, Zhongni Hou, Junshu Sun +5LLM Post-Training Fine-Tuning
State Key Lab. of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy Sciences · Meituan
Going Beyond State-Reaching: Learning Abstractions for Intrinsically Motivated Option Discovery
Sep 29, 2026Akhil Bagaria, Anita De Mello Koch, George KonidarisHierarchical RL RL Exploration
Brown University
X-Tree: Tokenizing Reusable Experience for Efficient Agent Generalization
Sep 26, 2026Sitao Cheng, Xunjian Yin, Zhiyuan Sun +4LLM Agent Skill Learning RL for LLM Agents
University of Waterloo · Duke University · National University of Singapore
What Does a ProcGen Generalization Gap Measure? Action Rules, Residual Entropy, and the Missing Random Floor
Sep 26, 2026Abhisek KeshariReinforcement Learning Generalization in RL
Independent researcher
Robust Successor Features
Sep 25, 2026Erik Nikulski, Yamen Habib, Vicenç Gomez +3Reinforcement Learning Generalization in RL
Department of Engineering Universitat Pompeu Fabra
Curriculum Learning with GNN-based Reinforcement Learning for Job Shop Scheduling
Sep 23, 2026Jayakrishnan K. Vasudevan, Jonathan Hoss, Noah KlarmannReinforcement Learning Curriculum RL
Rosenheim University of Applied Sciences
Categorical Internalisation of Environmental Groupoids for Generalisable POMDP Solving
Sep 23, 2026Ben Opperman, Eduardo Alonso, Esther MondragónReinforcement Learning Generalization in RL
City St George’s, University of London
Deep Reinforcement Learning on Item-Compatibility Graphs for One-Dimensional Bin Packing
Sep 21, 2026M. Aslı AydınCombinatorial Optimization RL for Combinatorial Optimization
Department of Management Engineering, Bahçeşehir University, Istanbul, Türkiye
The Free Inference Dimension: Complexity Measure for Zero-Collision Navigation under Hypothesis Mixtures
Sep 15, 2026Luiz Carlos Castro Guedes, Edward Hermann HaeuslerReinforcement Learning Meta-Reinforcement Learning
Departamento de Informática – PUC-Rio, Brazil · Seção de Engenharia de Computação – Instituto Militar de Engenharia, Brazil
Adaptive Chemotherapy Control under Tumor Heterogeneity via Reinforcement Learning
Sep 10, 2026Bereket Sitotaw Kidane, Md Samiul Haque Motayed, Shuo WangRL Control Reinforcement Learning
Department of Mechanical and Aerospace Engineering, University of Texas at Arlington, Arlington, TX 76019, USA
Efficient Exploration Is Enough
Sep 7, 2026Mikel Malagón, Jon Vadillo, Josu Ceberio +2Intrinsic Reward Methods RL Exploration
University of the Basque Country (UPV/EHU) · University of Alberta · Alberta Machine Intelligence Institute +2
Do Quantum AIs Dream in Paths? Path-Integral Slow Thinking through Grover Interference
Sep 5, 2026Xiansheng Cai, Xiu-Hao Deng, Kun ChenQuantum RL Generalization in RL
Institute of Theoretical Physics, Chinese Academy of Sciences, Beijing 100190, China · Shenzhen International Quantum Academy, Shenzhen 518048, China · Shenzhen Branch, Hefei National Laboratory, Shenzhen 518048, China
Uncertainty-Driven Replay Memory for Reinforcement Learning
Aug 30, 2026Sheeraja Rajakrishnan, Alexander G. Ororbia, Travis Desell +1Reinforcement Learning Uncertainty Estimation in RL
Rochester Institute of Technology
Active Curriculum Refinement for Reinforcement Learning
Aug 26, 2026Zhenya Liu, Yuxin ChenCurriculum RL Curriculum Learning
The University of Chicago, Chicago IL, USA.
Learning Generalizable Behaviors for Terminal Agents
Aug 23, 2026Yihang Yao, Bo Pang, Xuan Phi Nguyen +3Terminal Agents Generalization in RL
Salesforce AI Research · Carnegie Mellon University
Dynamic Context Scheduling: Learning Beyond the Static Universe
Aug 21, 2026Martin Mráz, André BiedenkappContextual RL Reinforcement Learning
University of Freiburg

Oct 21, 2026J/K move · Enter open · S save

Harness Scaling for Generalizable Agentic Reinforcement Learning

Fei Bai, Huatong Song, Chuan Hao, Ran Tao+4

Gaoling School of Artificial Intelligence, Renmin University of China · IQuest Research · Beihang University

Agentic RL LLM Agent Harnesses Generalization in RL

Open paper AskPDF arXiv

Agent harnesses have become a central component of modern agentic systems, shaping how models use tools, manage context, and coordinate long-horizon interaction. Realizing the benefits of these harnesses depends on how effectively the underlying model can make use of them, motivating reinforcement learning directly through the harness. While recent efforts have explored harness diversification to reduce specialization to particular execution settings, how to systematically construct and organize harness diversity for cross-harness generalization remains insufficiently understood. This work studies Harness Scaling as a distinct scaling dimension for agentic RL, focusing on structured variation in how policies execute and interact with the same underlying task and environment. Concretely, we construct a factorized harness space, select a compact training subset with balanced factorial coverage, and organize cross-harness training through locality-aware task assignment. Across multiple benchmarks, model scales, and RL algorithms, Harness Scaling consistently improves generalization to unseen harness combinations. Moreover, policies trained through structured white-box harness variation transfer effectively to unseen external black-box harnesses, outperforming policies trained through a single complex black-box harness. These results establish harness diversity as a meaningful scaling dimension for agentic RL and highlight the importance of how this diversity is organized for cross-harness generalization.