Reproduce “pico-WAM: Evaluating Counterfactual Prediction and Planning Error in a Compact World Action Model” 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
Keivalya Pandya
The pico-WAM links a policy, action-conditioned dynamics, and physical-state readout through a shared representation. It tests counterfactual simulator branches and search choices in supplied-state Push-T. Adding branches widens restart’s coverage advantage over warm-starting from 22.5 to 27.3 percentage points. Simulator scoring raises coverage by 12.2 points with learned proposals retained; these findings are limited to the tested setup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.