Evolving harnesses for various compute budgets
yassineyousfi_ · x · 2026-08-18
Inspired by work on Harness Engineering and Test-Time Compute, the author ran an experiment shifting the objective from pass rate to score = f(price). The goal is to evolve the best harness for various budgets, not just for infinite compute.
Related event: Experiments evolve AI harnesses to optimize inference cost(3 posts)→
More from Research
- GetPrescence models medicine as a game, building the first AI system to play for health — RishabJainK · 2026-08-19
- Nature paper: Epigenetic drift in microglia drives aging — rand_longevity · 2026-08-19
- NeurIPS Competition SimulacraBench: AI Survey Prediction — soumitrashukla9 · 2026-08-19
- AI for Science: Estimating Bacteria Size with Code — johnowhitaker · 2026-08-19
- UAI 2026 paper: identifiability metrics show systematic false positives in interpretability evals — RexDouglass · 2026-08-19
- GenBio AI releases AIDO Cell preview: one world model simulating the whole human cell — AllThingsApx · 2026-08-19