<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> This repository is a curated bibliography of research papers on In-Context Reinforcement Learning (ICRL). It addresses the challenge of tracking the rapidly advancing field by consolidating key publications, serving researchers and practitioners by providing a centralized, up-to-date resource for understanding ICRL frontiers.

How It Works

<2-4 sentences on core approach / design (key algorithms, models, data flow, or architectural choices) and why this approach is advantageous or novel.> The project functions as a living, community-driven list of ICRL research. Papers are organized chronologically, with direct links to their sources. This approach offers a consolidated, easily navigable overview of the academic landscape, facilitating discovery and synthesis of current research trends without manual aggregation.

Quick Start & Requirements

This section is omitted as the repository is a list of papers, not a runnable project.

Highlighted Details

• Research spans from pre-2023 to 2025, indicating sustained and growing academic interest.
• A prominent theme is the integration of Large Language Models (LLMs) into ICRL, exploring their use as RL agents, for imitation learning, and for implementing core RL algorithms.
• Key areas include meta-reinforcement learning, transformer architectures, and the development of specialized benchmarks and frameworks (e.g., OmniRL, LMAct) for evaluating ICRL capabilities.
• Notable research directions encompass generalization to novel sequential decision-making tasks, advanced exploration strategies, and adaptive agent development.

Maintenance & Community

Curated by dunnolab, the repository actively encourages community contributions via Pull Requests for new papers and resources, signaling ongoing maintenance and collaborative development.

Licensing & Compatibility

The README does not specify a software license for the repository itself. Users should verify the licensing terms of individual research papers and any associated code or datasets.

Limitations & Caveats

<1-3 sentences on caveats: unsupported platforms, missing features, alpha status, known bugs, breaking changes, bus factor, deprecation, etc. Avoid vague non-statements and judgments.> This is a bibliographic resource, not an executable framework or tool. Users must independently source, implement, and validate the research. The absence of a repository license may present compatibility challenges for integration or redistribution.