Reproduce “STRATA: A Unified Framework for Task Assignments in Large Teams of Heterogeneous 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

Harish Ravichandar, Kenneth Shaw, Sonia Chernova

STRATA assigns agents across tasks by matching capabilities to task requirements. It groups similar agents into species, models trait values and variation, and calculates moves over a task graph. In simulation, it converged in 79/100 exact-matching runs and 85/100 minimum-matching runs, versus 10/100 and 16/100 for an adapted binary baseline. In a game comparison, its advantage was not considerable.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.