Overview of RL Training

This section provides step-by-step guides for training all supported RL policies on OpenTau.

RL Training Guides:

• RECAP Training on pi0 policy
  • Introduction
  • Stage 1: SFT Training the VLA policy on whole libero dataset till convergence.
  • Stage 2: Fine-tuning the value function on whole libero dataset till convergence.
  • Stage 3: Offline RL training