Modal

Join us for a webinar with Joy Liu from the Modal Training Team. We’ll explore how engineers on Modal run inference, training workloads, and sandboxed environments across the entire RL pipeline. We'll dive into:

*   Taking advantage of multi-node training and Flash proxy for high-performance inference on Modal
*   A case study of RL fine-tuning of Qwen3-4B using LLM-as-a-judge reward signal
*   Scaffolding [SLIME](https://github.com/THUDM/slime) and other RL frameworks effectively on Modal
