Circuit Insights: Towards Interpretability Beyond Activations
This addresses the need for more systematic and scalable circuit analysis in explainable AI, though it builds incrementally on transcoder-based foundations.
The paper tackles the problem of limited scalability and interaction analysis in neural network interpretability by proposing WeightLens and CircuitLens, which interpret features from learned weights and capture circuit-level interactions, respectively, matching or exceeding existing methods on context-independent features.
The fields of explainable AI and mechanistic interpretability aim to uncover the internal structure of neural networks, with circuit discovery as a central tool for understanding model computations. Existing approaches, however, rely on manual inspection and remain limited to toy tasks. Automated interpretability offers scalability by analyzing isolated features and their activations, but it often misses interactions between features and depends strongly on external LLMs and dataset quality. Transcoders have recently made it possible to separate feature attributions into input-dependent and input-invariant components, providing a foundation for more systematic circuit analysis. Building on this, we propose WeightLens and CircuitLens, two complementary methods that go beyond activation-based analysis. WeightLens interprets features directly from their learned weights, removing the need for explainer models or datasets while matching or exceeding the performance of existing methods on context-independent features. CircuitLens captures how feature activations arise from interactions between components, revealing circuit-level dynamics that activation-only approaches cannot identify. Together, these methods increase interpretability robustness and enhance scalable mechanistic analysis of circuits while maintaining efficiency and quality.