RSI-Anything's lab-planning score reportedly more than doubles across five policy versions

A post about RSI-Anything reports gains across code repair, lab planning and reaction optimization. It says a predictive world model chooses where to spend costly real feedback, which updates both the policy and model. Lab-planning scores rose from 21.80% to 46.01% across five policy versions; in reaction optimization, mean best yield reached 96.15%, versus 82.16% for Bayesian optimization with the same 40 experiments.