Architecture, Not Scale: Circuit Localization in Large Language Models

• URL: http://arxiv.org/abs/2605.08853v1
• Date: Sat, 09 May 2026 10:05:22 GMT
• Title: Architecture, Not Scale: Circuit Localization in Large Language Models
• Abstract summary: We show that grouped query attention produces circuits far more concentrated and mechanistically stable than standard multi-head attention at comparable scales.<n>The same concentration pattern holds across indirect object identification, induction heads, and factual recall.<n>These findings suggest that some architectural choices make large models more tractable to study.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Mechanistic interpretability assumes that circuit analysis becomes harder as models scale. We challenge this assumption by showing that the attention architecture matters more than parameter count. Studying three circuit types across Pythia and Qwen2.5, we find that grouped query attention produces circuits that are far more concentrated and mechanistically stable than standard multi-head attention at comparable scales. The same concentration pattern holds across indirect object identification, induction heads, and factual recall. Within a single architecture family (Qwen2.5), factual recall circuits undergo a discrete phase transition above a critical scale, collapsing to a single bottleneck rather than degrading gradually. These findings suggest that some architectural choices make large models more tractable to study and that interpretability difficulty is not a fixed consequence of model size.

Related papers

• Differentiable Faithfulness Alignment for Cross-Model Circuit Transfer [73.0825876244911]
  We introduce DFA, a framework that transfers circuit information from a smaller source model to a larger target model.<n>We evaluate DFA on Llama-3 and Qwen-2.5 across six tasks spanning factual retrieval, multiple-choice reasoning, and arithmetic.<n>DFA consistently outperforms simple baselines and, in some settings, recovers target-model circuits with faithfulness comparable to or stronger than direct attribution.
  arXiv  Detail & Related papers  (2026-04-27T10:49:10Z)
• Quantifying LLM Attention-Head Stability: Implications for Circuit Universality [2.6132365604376404]
  We study stability across-refits in increasingly complex transformer language models of various sizes.<n>Middle-layer heads are the least stable yet the most representationally distinct.<n>Deeper models exhibit stronger mid-depth divergence.
  arXiv  Detail & Related papers  (2026-02-17T23:30:59Z)
• Dynamics Within Latent Chain-of-Thought: An Empirical Study of Causal Structure [58.89643769707751]
  We study latent chain-of-thought as a manipulable causal process in representation space.<n>We find that latent-step budgets behave less like homogeneous extra depth and more like staged functionality with non-local routing.<n>These results motivate mode-conditional and stability-aware analyses as more reliable tools for interpreting and improving latent reasoning systems.
  arXiv  Detail & Related papers  (2026-02-09T15:25:12Z)
• How Attention Sinks Emerge in Large Language Models: An Interpretability Perspective [67.08306259095778]
  We identify a simple mechanism that enables the model to recognize token at position zero and induce an attention sink within two transformer blocks.<n>By analyzing training traces from a 30B A3B MoE model trained from scratch, we find that this mechanism emerges early in training and becomes increasingly concentrated in the first two layers.
  arXiv  Detail & Related papers  (2026-02-04T11:10:34Z)
• Right Looks, Wrong Reasons: Compositional Fidelity in Text-to-Image Generation [34.89831673834612]
  This survey investigates this breakdown across three core primitives-negation, counting, and spatial relations.<n>We trace this failure to three key factors. First, training data show a near-total absence of explicit negations.<n>We conclude, will require fundamental advances in representation and reasoning rather than incremental adjustments to existing architectures.
  arXiv  Detail & Related papers  (2025-11-13T09:49:21Z)
• Scaling Laws and Symmetry, Evidence from Neural Force Fields [14.109815254143205]
  We show a clear power-law scaling behaviour with respect to data, parameters and compute with architecture-dependent exponents''<n>In particular, we observe that equivariant architectures, which leverage task symmetry, scale better than non-equivariant models.<n>Our analysis also suggests that for compute-optimal training, the data and model sizes should scale in tandem regardless of the architecture.
  arXiv  Detail & Related papers  (2025-10-10T18:22:00Z)
• Why Can't Transformers Learn Multiplication? Reverse-Engineering Reveals Long-Range Dependency Pitfalls [54.57326125204404]
  Language models are increasingly capable, yet still fail at a seemingly simple task of multi-digit multiplication.<n>We study why, by reverse-engineering a model that successfully learns multiplication via emphimplicit chain-of-thought'
  arXiv  Detail & Related papers  (2025-09-30T19:03:26Z)
• Scaling Laws vs Model Architectures: How does Inductive Bias Influence Scaling? [91.78878523252897]
  This paper conducts a systematic study of scaling behaviour of ten diverse model architectures. We show that architecture is an important consideration when performing scaling and the best performing model can fluctuate at different scales.
  arXiv  Detail & Related papers  (2022-07-21T15:50:22Z)
• Systematic Evaluation of Causal Discovery in Visual Model Based Reinforcement Learning [76.00395335702572]
  A central goal for AI and causality is the joint discovery of abstract representations and causal structure. Existing environments for studying causal induction are poorly suited for this objective because they have complicated task-specific causal graphs. In this work, our goal is to facilitate research in learning representations of high-level variables as well as causal structures among them.
  arXiv  Detail & Related papers  (2021-07-02T05:44:56Z)

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

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.