Excited to share SkyRL-SQL, a simple yet effective multi-turn RL pipeline for training LLMs to generate and refine SQL through real database feedback. Rather than one-shot generation, models explore unfamiliar schemas, issue trial queries, reflect on results, and iteratively

The SkyRL roadmap is live! Our focus is on building the easiest-to-use high-performance RL framework for agents. We'd love your ideas, feedback, or code to guide the project: github.com/NovaSky-AI/Sky…

🔥 At ICML 2025, we’re delighted to introduce BFCL V4 Agentic. As function-calling (also called tool-calling) forms the bed-rock of Agentic systems, BFCL V4 Agentic benchmark focuses on tool-calling in real-world agentic settings — including: 🔍 Web search with multi-hop

By the way, many of you SkyRL users have been hacking on custom advantage estimation functions but had to modify SkyRL internals to get it to work. Just added a registry for custom adv. estimators to make this much easier: github.com/NovaSky-AI/Sky…

🥇 Happy to share that our work on SkyRL won the gold medal in "Reasoning & Planning Research Track" of the UC Berkeley RDI AgentX Competition. Thanks for the great event! rdi.berkeley.edu/agentx/