Path patching provides a method to express and quantitatively test hypotheses that neural network behaviors are localized to sets of paths.
Finding alignments between interpretable causal variables and distributed neural representations
7 Pith papers cite this work, alongside 9 external citations. Polarity classification is still indexing.
7
Pith papers citing it
9
external citations · external index
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
unclear 1representative citing papers
The paper introduces compositional interpretability as a category-theoretic framework that casts mechanistic explanations as commuting syntactic-semantic mappings optimized under faithfulness and complexity constraints derived from minimum description length.
Local linearity of LLM layers enables LQR-based closed-loop activation steering with theoretical tracking guarantees.
A latent mediation framework with sparse autoencoders enables non-additive token-level influence attribution in LLMs by learning orthogonal features and back-propagating attributions.
Perturbation probing identifies tiny sets of FFN neurons that control refusal templates and language routing in LLMs, enabling precise ablations and directional interventions that alter behavior on benchmarks while preserving safety.
Overthinking in medical QA is linearly decodable at 71.6% accuracy yet fixed residual-stream steering yields no correction across 29 configurations, while enabling selective abstention with AUROC 0.610.
Varying evaluation metrics and corruption methods in activation patching produces different localization and circuit discovery outcomes in language models, leading to recommendations for preferred practices.
citing papers explorer
-
Localizing Model Behavior with Path Patching
Path patching provides a method to express and quantitatively test hypotheses that neural network behaviors are localized to sets of paths.
-
From Mechanistic to Compositional Interpretability
The paper introduces compositional interpretability as a category-theoretic framework that casts mechanistic explanations as commuting syntactic-semantic mappings optimized under faithfulness and complexity constraints derived from minimum description length.
-
Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal Control
Local linearity of LLM layers enables LQR-based closed-loop activation steering with theoretical tracking guarantees.
-
Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces
A latent mediation framework with sparse autoencoders enables non-additive token-level influence attribution in LLMs by learning orthogonal features and back-propagating attributions.
-
Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs
Perturbation probing identifies tiny sets of FFN neurons that control refusal templates and language routing in LLMs, enabling precise ablations and directional interventions that alter behavior on benchmarks while preserving safety.
-
Decodable but Not Corrected by Fixed Residual-Stream Linear Steering: Evidence from Medical LLM Failure Regimes
Overthinking in medical QA is linearly decodable at 71.6% accuracy yet fixed residual-stream steering yields no correction across 29 configurations, while enabling selective abstention with AUROC 0.610.
-
Towards Best Practices of Activation Patching in Language Models: Metrics and Methods
Varying evaluation metrics and corruption methods in activation patching produces different localization and circuit discovery outcomes in language models, leading to recommendations for preferred practices.