Towards Direct Evaluation of Harness Optimizers via Priority Ranking

TL;DR

This paper introduces a low-cost, direct evaluation method called priority ranking for harness optimizers, which predicts their multi-step optimization performance without costly rollouts.

Contribution

It proposes a novel priority ranking approach to evaluate harness optimizers directly, validated by a diverse set of scenarios and correlating well with actual optimization success.

Findings

Priority ranking correlates with multi-step optimization performance.

The method enables cost-effective, step-level evaluation of harness optimizers.

A collection of 182 human-verified scenarios supports evaluation across domains.

Abstract

Harness optimization enables automated agent creation by having an optimizer agent iteratively update the harness of target agents. Despite its success, current studies evaluate optimizers solely by observing target agents' performance gains. This indirect end-improvement evaluation neglects optimizers' actions at intermediate steps, which are often erroneous and hinder agent performance. Therefore, it is unclear whether harness optimization is driven by optimizers' informed update actions or simply trial-and-error. This necessitates direct evaluation of harness optimizers. However, evaluating harness optimizers directly is non-trivial and costly due to the lack of oracle harnesses. To address this, we present a simple, low-cost design to directly evaluate them, namely priority ranking. By asking harness optimizers to rank components (e.g., tools) in a given harness by their potential to improve/hinder agent performance when updated, our design quantifies optimizer ability at the step level without expensive rollouts or manual examination. More importantly, optimizers' ranking performance correlates with their ability to improve agents in actual multi-step harness optimization, establishing priority ranking as a reliable predictor of optimization ability. Priority ranking is enabled by Shor, a collection of 182 human-verified optimization scenarios spanning across domains, designs, and time stages. Codes and data can be found at https://github.com/k59118/Harness_Optimizer_Evaluation.