Circuits in Language Models

Follow this topic Ask about this topic

Circuits in language models are groups of attention heads, MLP neurons, and learned features whose interactions support behaviors such as induction, factual recall, arithmetic, refusal, and multilingual routing. Researchers locate them with activation patching and ablation, attribution graphs, and sparse autoencoders or transcoders, then test their causal roles on controlled tasks such as indirect object identification and greater-than. Some computations appear compact and reusable, while others are distributed, context-dependent, or reorganized by architecture and post-training. A central challenge is showing that a discovered circuit is causally faithful and task-specific rather than an artifact of prompts, analysis choices, or intervention side effects.

Latest papers

122 papers27 papers in the last four weeks, up 42% on the four weeks before. 0.2% of all new papers.

Quantifying LLM Attention-Head Stability: Implications for Circuit Universality
Feb 17, 2026Karan Bali, Jack Stanley, Praneet Suresh +1Transformer Interpretability Attention Head Analysis
Mila - Quebec Artificial Intelligence Institute, Montreal, Canada · McGill University, Montreal, Canada
Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models
Jan 21, 2026Injin Kong, Hyoungjoon Lee, Yohan JoMasked Diffusion Models Autoregressive Diffusion
Graduate School of Data Science, Seoul National University · Department of Biosystems & Biomaterials Science and Engineering, Seoul National University

Feb 17, 2026J/K move · Enter open · S save

Quantifying LLM Attention-Head Stability: Implications for Circuit Universality

Karan Bali, Jack Stanley, Praneet Suresh, Danilo Bzdok

Mila - Quebec Artificial Intelligence Institute, Montreal, Canada · McGill University, Montreal, Canada

Transformer Interpretability Attention Head Analysis Circuits in Language Models

Open paper AskPDF arXiv

In mechanistic interpretability, recent work scrutinizes transformer "circuits" - sparse, mono or multi layer sub computations, that may reflect human understandable functions. Yet, these network circuits are rarely acid-tested for their stability across different instances of the same deep learning architecture. Without this, it remains unclear whether reported circuits emerge universally across labs or turn out to be idiosyncratic to a particular estimation instance, potentially limiting confidence in safety-critical settings. Here, we systematically study stability across-refits in increasingly complex transformer language models of various sizes. We quantify, layer by layer, how similarly attention heads learn representations across independently initialized training runs. Our rigorous experiments show that (1) middle-layer heads are the least stable yet the most representationally distinct; (2) deeper models exhibit stronger mid-depth divergence; (3) unstable heads in deeper layers become more functionally important than their peers from the same layer; (4) applying weight decay optimization substantially improves attention-head stability across random model initializations; and (5) the residual stream is comparatively stable. Our findings establish the cross-instance robustness of circuits as an essential yet underappreciated prerequisite for scalable oversight, drawing contours around possible white-box monitorability of AI systems.

Latest papers

Quantifying LLM Attention-Head Stability: Implications for Circuit Universality

Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models