Systems Reading Group with Resolve AI - Scaling RL Compute

​Join us for June's Systems Reading Group with Resolve AI!

​This session will be led by Sean Bell, who leads RL research at Resolve AI Labs. Previously, he led all LLM pre-training data at Meta Superintelligence Labs, and was the founder of GrokStyle, which was acquired by Facebook.

​We'll dive into the paper "The Art of Scaling Reinforcement Learning Compute for LLMs"(https://arxiv.org/pdf/2510.13786), and discuss questions like "When doing RL yourself, how do you actually set all the parameters?  What matters and what doesn't?"

​We'll also discuss this in the context of Resolve's post-training plans, how they do RCA, and how they build custom models to get better speed-accuracy tradeoffs.

​Reading the paper beforehand is encouraged, but not required for this session.

​Come by Resolve's office on 6/24 for one of our most anticipated sessions!