Related papers: How Much Do Circuits Tell Us? Measuring the Consistency and Specificity of Language Model Circuits

How Much Do Circuits Tell Us? Measuring the Consistency and Specificity of Language Model Circuits

URL: http://arxiv.org/abs/2605.08348v1
Date: Fri, 08 May 2026 18:01:03 GMT
Title: How Much Do Circuits Tell Us? Measuring the Consistency and Specificity of Language Model Circuits
Abstract summary: We measure circuit reuse, the proportion of components shared across per-example circuits within a task, and investigate two less-studied properties of this consistency.<n>We find that within-task reuse is high and that shared components are necessary for task performance, with ablations causing up to $sim$100% relative accuracy drops.<n>However, circuits turn out not to be task-specific: ablating one task's circuit damages another task's performance about as much as that task's own circuit does.
Score: 4.991808275998526
License: http://creativecommons.org/licenses/by/4.0/
Abstract: The circuits framework in mechanistic interpretability aims to identify causally important sparse subgraphs of model components, typically evaluated by measuring necessity and sufficiency. We measure circuit reuse, the proportion of components shared across per-example circuits within a task, and investigate two less-studied properties of this: consistency, the recurrence of components within a task, and specificity, their uniqueness to a task. Using edge attribution patching across six tasks and seven models, we find that within-task reuse is high and that shared components are necessary for task performance, with ablations causing up to $\sim$100% relative accuracy drops. However, circuits turn out not to be task-specific: ablating one task's circuit damages another task's performance about as much as that task's own circuit does. We discover that this is due to substantial overlap between circuits across tasks, which are causally important for performance. Some circuits do contain a smaller set of task-specific components, but these account for only a modest portion of circuit performance. Overall, our findings suggest that while circuit discovery at the level of attention heads and MLP layers identifies important components, their lack of task-specificity raises questions about the degree to which circuits can support targeted understanding and intervention on model behavior.

Related papers

S^3martCirc: Self-supervised Smart Circuit Discovery [44.37905543632994]
Large Language Models (LLMs) have demonstrated remarkable performance across diverse tasks.<n>Despite these capabilities, their black-box nature obscures internal decision-making processes.<n>Mechanistic interpretability aims to address this by reverse-engineering neural networks into human-understandable algorithms.
arXiv Detail & Related papers (2026-09-01T05:37:42Z)
IBCircuit: Towards Holistic Circuit Discovery with Information Bottleneck [39.572087058128645]
We propose an end-to-end approach based on the principle of Information Bottleneck, called IBCircuit, to identify informative circuits holistically.<n> IBCircuit is an optimization framework for holistic circuit discovery and can be applied to any given task without tediously corrupted activation design.
arXiv Detail & Related papers (2026-02-26T03:33:35Z)
Purifying Task Vectors in Knowledge-Aware Subspace for Model Merging [83.5273168208788]
Model merging aims to integrate task-specific abilities from individually fine-tuned models into a single model without extra training.<n>The merged model often suffers from notable performance degradation due to the conflicts caused by task-irrelevant redundancy in task vectors.<n>We propose Purifying TAsk Vectors (PAVE) in knowledge-aware subspace to overcome these challenges.
arXiv Detail & Related papers (2025-10-16T14:02:57Z)
Circuit Compositions: Exploring Modular Structures in Transformer-Based Language Models [22.89563355840371]
We study the modularity of neural networks by analyzing circuits for highly compositional subtasks within a language model.<n>Our results indicate that functionally similar circuits exhibit both notable node overlap and cross-task faithfulness.
arXiv Detail & Related papers (2024-10-02T11:36:45Z)
Transformer Circuit Faithfulness Metrics are not Robust [0.04260910081285213]
We measure circuit 'faithfulness' by ablating portions of the model's computation. We conclude that existing circuit faithfulness scores reflect both the methodological choices of researchers as well as the actual components of the circuit. The ultimate goal of mechanistic interpretability work is to understand neural networks, so we emphasize the need for more clarity in the precise claims being made about circuits.
arXiv Detail & Related papers (2024-07-11T17:59:00Z)
A Unified Causal View of Instruction Tuning [76.1000380429553]
We develop a meta Structural Causal Model (meta-SCM) to integrate different NLP tasks under a single causal structure of the data. Key idea is to learn task-required causal factors and only use those to make predictions for a given task.
arXiv Detail & Related papers (2024-02-09T07:12:56Z)
Circuit Component Reuse Across Tasks in Transformer Language Models [32.2976613483151]
We present evidence that insights can indeed generalize across tasks. We show that the process underlying both tasks is functionally very similar, and contains about a 78% overlap in in-circuit attention heads. Our results provide evidence that it may yet be possible to explain large language models' behavior in terms of a relatively small number of interpretable task-general algorithmic building blocks and computational components.
arXiv Detail & Related papers (2023-10-12T22:12:28Z)
Fast Inference and Transfer of Compositional Task Structures for Few-shot Task Generalization [101.72755769194677]
We formulate it as a few-shot reinforcement learning problem where a task is characterized by a subtask graph. Our multi-task subtask graph inferencer (MTSGI) first infers the common high-level task structure in terms of the subtask graph from the training tasks. Our experiment results on 2D grid-world and complex web navigation domains show that the proposed method can learn and leverage the common underlying structure of the tasks for faster adaptation to the unseen tasks.
arXiv Detail & Related papers (2022-05-25T10:44:25Z)
Distribution Matching for Heterogeneous Multi-Task Learning: a Large-scale Face Study [75.42182503265056]
Multi-Task Learning has emerged as a methodology in which multiple tasks are jointly learned by a shared learning algorithm. We deal with heterogeneous MTL, simultaneously addressing detection, classification & regression problems. We build FaceBehaviorNet, the first framework for large-scale face analysis, by jointly learning all facial behavior tasks.
arXiv Detail & Related papers (2021-05-08T22:26:52Z)
Multi-task Supervised Learning via Cross-learning [102.64082402388192]
We consider a problem known as multi-task learning, consisting of fitting a set of regression functions intended for solving different tasks. In our novel formulation, we couple the parameters of these functions, so that they learn in their task specific domains while staying close to each other. This facilitates cross-fertilization in which data collected across different domains help improving the learning performance at each other task.
arXiv Detail & Related papers (2020-10-24T21:35:57Z)
Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference [75.95287293847697]
Two common challenges in developing multi-task models are often overlooked in literature. First, enabling the model to be inherently incremental, continuously incorporating information from new tasks without forgetting the previously learned ones (incremental learning) Second, eliminating adverse interactions amongst tasks, which has been shown to significantly degrade the single-task performance in a multi-task setup (task interference)
arXiv Detail & Related papers (2020-07-24T14:44:46Z)

This list is automatically generated from the titles and abstracts of the papers in this site.