Don't Let Systems Swallow the Algorithm
Why RL for large models needs separable algorithm, rollout, and orchestration layers.
Read Post
FeynRL
Understand What You Build.
A post-training framework for large models — from new objectives to new rollout systems, where systems stay systems and algorithms stay algorithms.
About
“What I cannot create, I do not understand.” — Richard Feynman
FeynRL (pronounced "FineRL") is an algorithm-first framework for post-training and fine-tuning large models. It supports supervised fine-tuning, preference learning, and reinforcement learning, and it is built for researchers and engineers who want to understand, modify, and develop new methods without fighting the infrastructure.
News
FeynRL Open-Sourced
The full codebase, methods, and example training recipes are now available on GitHub.
Open GitHub
Experiments
Training curves, benchmark summaries, and release metrics now live on a dedicated experiments page and will continue to expand with new runs.