Skip to main content

World models for agent training

Tracks claims about using learned world models to train or improve agents through simulated or imagined experience.

Entries in This Trail

  1. Dreamer 4 becomes first reported agent to obtain Minecraft diamonds using only offline training data

    Google DeepMind researchers report that Dreamer 4 obtained diamonds in 0.7% of 1,000 one-hour Minecraft evaluations after learning from a fixed 2,541-hour gameplay dataset and improving its policy through reinforcement learning entirely inside a learned world model, without online environment interaction during training.

    Claim: SupportedEvidence: StrongReview: Stable