This folder contains the supplementary material for submission "Towards Universal Visual Reward and Representation via Value-Implicit Pre-Training"

1. code.zip contains code for training a VIP model from scratch on custom video datasets.

2. videos.pptx contains policy rollout videos for our real-world robot experiments.

3. project website: https://sites.google.com/view/rl-vip; the project website provides more compact viewing of the videos in videos.pptx