1
Review Documentation
Visit https://verl.readthedocs.io to understand the framework overview and agent framework details.
2
Integrate with LLM Infrastructure
Use verl's modular APIs to connect with existing LLM frameworks such as PyTorch FSDP or vLLM.
3
Build RL Dataflows
Construct reinforcement learning dataflows using the hybrid programming model with minimal code.
4
Configure Device Mapping
Set up flexible GPU device mapping and parallelism across clusters to optimize training performance.
5
Train with HuggingFace Models
Leverage ready integration with HuggingFace models for reinforcement learning training.