BlackboxNLP-2025 MIB Shared Task: Improving Circuit Faithfulness via Better Edge Selection
This work addresses a key bottleneck in interpretability for researchers, though it appears incremental as it builds directly on the existing MIB framework.
The paper tackles the challenge of circuit discovery in mechanistic interpretability by proposing three improvements to edge selection, resulting in more faithful circuits that outperform prior approaches across multiple MIB tasks and models.
One of the main challenges in mechanistic interpretability is circuit discovery, determining which parts of a model perform a given task. We build on the Mechanistic Interpretability Benchmark (MIB) and propose three key improvements to circuit discovery. First, we use bootstrapping to identify edges with consistent attribution scores. Second, we introduce a simple ratio-based selection strategy to prioritize strong positive-scoring edges, balancing performance and faithfulness. Third, we replace the standard greedy selection with an integer linear programming formulation. Our methods yield more faithful circuits and outperform prior approaches across multiple MIB tasks and models. Our code is available at: https://github.com/technion-cs-nlp/MIB-Shared-Task.