Reproduce “Building World Models from Intrinsic Motivation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Reece Keller, Declan Kutscher, Xaq Pitkow, Aran Nayebi
The authors compare 24 intrinsic objectives and three world-model representation strategies in a Dreamer-based framework for reward-free pretraining. They evaluate models on downstream control and few-shot adaptation. Curiosity with observation reconstruction has the strongest aggregate family profile. In Craftax vision, NGU with DreamerV3 reaches 15.8 achievements versus 5.9 from scratch after 500K task steps. This is not a universal result.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.