Heading to NeurIPS 2025🥳. Check out our new work on dissecting math reasoning for RL! If you’re exploring anything in LLM agents, RL, or AI for game — I’d love to chat. We are also hiring interns on LLMs and RL. DM me if interested @SFResearch
#NeurIPS2025 #LLM
Ever wondered:
• Does RL actually improve reasoning?
• How do RL-tuned models use auxiliary information and what’s the most effective way to provide hints?
• Are "hard problems" with no positive advantages still useful for RL training?
I’ll be answering these questions!
📍
There’s a new version of this post
