Attribution Patching Outperforms Automated Circuit Discovery
Reviewed by Pithpith:IWEM3X4Dopen to challenge →
read the original abstract
Automated interpretability research has recently attracted attention as a potential research direction that could scale explanations of neural network behavior to large models. Existing automated circuit discovery work applies activation patching to identify subnetworks responsible for solving specific tasks (circuits). In this work, we show that a simple method based on attribution patching outperforms all existing methods while requiring just two forward passes and a backward pass. We apply a linear approximation to activation patching to estimate the importance of each edge in the computational subgraph. Using this approximation, we prune the least important edges of the network. We survey the performance and limitations of this method, finding that averaged over all tasks our method has greater AUC from circuit recovery than other methods.
This paper has not been read by Pith yet.
Forward citations
Cited by 18 Pith papers
-
WriteSAE: Sparse Autoencoders for Recurrent State
WriteSAE is the first sparse autoencoder that factors decoder atoms into the native d_k x d_v cache write shape of recurrent models and supplies a closed-form per-token logit shift for atom substitution.
-
WriteSAE: Sparse Autoencoders for Recurrent State
WriteSAE introduces sparse autoencoders with rank-1 matrix atoms for recurrent state updates, allowing replacement tests that outperform deletion on 92.4% of positions and a formula predicting logit changes with R²=0.98.
-
WriteSAE: Sparse Autoencoders for Recurrent State
WriteSAE decomposes recurrent model cache writes into substitutable atoms with a closed-form logit shift, achieving high substitution success and targeted behavioral installs on models like Qwen3.5 and Mamba-2.
-
Interaction Locality in Hierarchical Recursive Reasoning
Interaction locality is introduced as a task-geometry-aware measurement framework showing that high-level states in recursive models write locally while recursive updates build broader structures on maze, Sudoku, ARC-...
-
WriteSAE: Sparse Autoencoders for Recurrent State
WriteSAE factors sparse autoencoder decoder atoms to the native d_k x d_v cache write shape in recurrent models, provides a closed-form logit shift, and demonstrates high success in atom substitution and behavioral ed...
-
Faithfulness to Refusal: A Causal Audit of Neuron Selectors
A causal audit via neuron-row zeroing shows attribution methods (LRP, IG) faithfully identify dispensable neurons and can install refusal behavior, while rank-stability proxies systematically miss selector failures.
-
Validating Causal Abstraction Metrics on Simulated Complex Systems
Authors create a benchmark across discrete/continuous and static/dynamical systems and introduce the Causal Abstraction Error (CAE) metric that reliably distinguishes valid from invalid causal abstractions when it inc...
-
Do Models Read What They Write? Causal Registers in Scratchpad Reasoning
State-writing models causally use edited scratchpad states in a controlled task at 80-91% accuracy on held-out examples, unlike final-answer-only and pretrained controls.
-
Quantifying the Agreement Between Data-Influence and Data-Similarity to Understand LLM Behavior
Data-similarity and data-influence produce significantly overlapping rankings of training documents for LLM outputs, with asymmetry allowing a favorable cost-accuracy trade-off.
-
Relational Rank Geometry in Transformers: Detecting and Steering Hidden-State Relation Frames
Transformer hidden states contain rank-indexed orientation signatures for true r-argument relations (r=3-6) that survive surface controls and can be patched to alter model outputs on relation tasks.
-
Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
Sparse autoencoders scaled to 34 million features on Claude 3 Sonnet yield interpretable, steerable representations of concrete and abstract concepts that generalize across languages and modalities.
-
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.
-
Instructions Shape Production of Language, not Processing
Instructions trigger a production-centered mechanism in language models, with task-specific information stable in input tokens but varying strongly in output tokens and correlating with behavior.
-
Dictionary-Aligned Concept Control for Safeguarding Multimodal LLMs
DACO curates a 15,000-concept dictionary from 400K image-caption pairs and uses it to initialize an SAE that enables granular, concept-specific steering of MLLM activations, raising safety scores on MM-SafetyBench and...
-
Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings
Circuit-based metrics from Vision Transformer internals provide better label-free proxies for generalization under distribution shift than existing methods like model confidence.
-
Instructions Shape Production of Language, not Processing
Instructions primarily shape the production stage of language models rather than the processing stage, with task-specific information and causal effects stronger in output tokens than input tokens.
-
Pruning Unsafe Tickets: A Resource-Efficient Framework for Safer and More Robust LLMs
Pruning removes 'unsafe tickets' from LLMs via gradient-free attribution, reducing harmful outputs and jailbreak vulnerability with minimal utility loss.
-
Attribution-Guided Pruning for Insight and Control: Circuit Discovery and Targeted Correction in Small-scale LLMs
Attribution-guided pruning with contrastive relevance identifies behavior-specific circuits in small LLMs and removes as little as 0.03-0.3% of components to reduce toxicity or repetition while preserving general performance.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.