Story

In brief

Reinforcement learning for LLMs is hard, so we built Modal Dojo — a library that gives you the building blocks, tuned defaults, and observability to get started with production-grade LLM training.