Agent S: an open agentic framework that uses computers like a human
From idea to production in just few lines: Graph-Based Programmable Neuro-Symbolic LM Framework - a production-first LM framework built with decade old Deep Learning best practices
Official code for "SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization"
Awesome In-Context RL: A curated list of In-Context Reinforcement Learning - - —
Official Implementation for "In-Context Reinforcement Learning for Variable Action Spaces"
XLand-100B: A Large-Scale Multi-Task Dataset for In-Context Reinforcement Learning - - — ICLR 2025
Vintix: Action Model via In-Context Reinforcement Learning - - — ICML 2025
Official Implementation for "In-Context Reinforcement Learning from Noise Distillation"
[NeurIPS 2025] Official codebase for T2MIR: Mixture-of-Experts Meets In-Context Reinforcement Learning.
Next-gen Foundation Model for Embodied AI
Benchmarking general decision-making with open & random worlds
A repo for enhancing spatial reasoning in VLMs using CoT and VoT prompting for 3D visual environments
Toy meta-RL environments for testing algorithms implementations
Explore the "frp_rl" repository to discover the Free Random Projection technique for in-context reinforcement learning. This project offers a JAX-based implementation that helps agents adapt and learn in diverse environments. 🐙🌟
Grouped-Tied Attention by Zadouri, Strauss, Dao (2025).
Source code for reproducing free random projection
Testing hypotheses of distillation of a learning algorithm on multiple environments
In-Context Reinforcement Learning for LLMs — no fine-tuning, no gradient updates.
Repository Switched To Xenoverse (https://github.com/FutureAGI/Xenoverse)