Reproduce “Learning Dynamics Models for Model Predictive Agents” 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
Michael Lutter, Leonard Hasenclever, Arunkumar Byravan, Gabriel Dulac-Arnold, Piotr Trochim, Nicolas Heess, Josh Merel, Yuval Tassa
The paper compares choices for training learned dynamics models used by model-predictive control (MPC). Feed-forward models trained on 500 episodes vary deterministic or stochastic predictions, training horizon, ensemble size, and input noise, then guide planning across five simulated tasks. Both model types support successful planning in four tasks, with rewards marginally below the simulator; both fail on humanoid stand.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.