Agent psychometrics: Task-level performance prediction in agentic coding benchmarks
This work addresses the problem for benchmark designers in AI coding, allowing them to calibrate task difficulty without costly evaluations, though it is incremental as it builds on existing IRT methods.
The paper tackles the challenge of predicting success or failure on individual tasks in agentic coding benchmarks, where current aggregate metrics obscure task diversity, by introducing a framework that augments Item Response Theory with task features and decomposes agent ability into LLM and scaffold components, enabling accurate predictions for unseen benchmarks and LLM-scaffold combinations.
As the focus in LLM-based coding shifts from static single-step code generation to multi-step agentic interaction with tools and environments, understanding which tasks will challenge agents and why becomes increasingly difficult. This is compounded by current practice: agent performance is typically measured by aggregate pass rates on benchmarks, but single-number metrics obscure the diversity of tasks within a benchmark. We present a framework for predicting success or failure on individual tasks tailored to the agentic coding regime. Our approach augments Item Response Theory (IRT) with rich features extracted from tasks, including issue statements, repository contexts, solutions, and test cases, and introduces a novel decomposition of agent ability into LLM and scaffold ability components. This parameterization enables us to aggregate evaluation data across heterogeneous leaderboards and accurately predict task-level performance for unseen benchmarks, as well as unseen LLM-scaffold combinations. Our methods have practical utility for benchmark designers, who can better calibrate the difficulty of their new tasks without running computationally expensive agent evaluations.