VeRL-Tune
Scalable HPO and evaluation pipeline for Agentic RL.
VeRL-Tune is an open-source Python framework designed to bridge the gap in the Agentic AI ecosystem by providing standardized hyperparameter optimization (HPO) and evaluation for the VeRL framework.
It provides a seamless bridge between VeRL and Ray Tune to enable scalable, multi-node HPO and packages a standardized evaluation module for Agentic RL/LLM reasoning.