Modeling Transformers as complex networks to analyze learning dynamics
On the map: Mechanistic interpretability → Knowledge and representations inside LLMs
Abstract (arXiv)
The process by which Large Language Models (LLMs) acquire complex capabilities during training remains a key open question in mechanistic interpretability. This project investigates whether these learning dynamics can be characterized through the lens of Complex Network Theory (CNT). I introduce a novel methodology to represent a Transformer-based LLM as a directed, weighted graph where nodes are the model's computational components (attention heads and MLPs) and edges represent causal influence, measured via an intervention-based ablation technique. By tracking the evolution of this component-graph across 143 training checkpoints of the Pythia-14M model on a canonical induction task, I analyze a suite of graph-theoretic metrics. The results reveal that the network's structure evolves through distinct phases of exploration, consolidation, and refinement. Specifically, I identify the emergence of a stable hierarchy of information spreader components and a dynamic set of information gatherer components, whose roles reconfigure at key learning junctures. This work demonstrates that a component-level network perspective offers a powerful macroscopic lens for visualizing and understanding the self-organizing principles that drive the formation of functional circuits in LLMs.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Knowledge and representations inside LLMs, one of 13 directions in Mechanistic interpretability. Builds on 5 of its 9 references that are in the map. First posted on arXiv 18 Sep 2025.
Builds on
5 of 9 references.
References · 5
- 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,786
- 2022In-context Learning and Induction Headscited by 964
- 2020Graph Structure of Neural Networkscited by 68
- 2021Structure and Performance of Fully Connected Neural Networks: Emerging Complex Network Propertiescited by 12
- 2021Characterizing Learning Dynamics of Deep Neural Networks via Complex Networkscited by 8
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025GraphGhost: Tracing Structures Behind Large Language Models
- 2026Invariant Learning Dynamics of Transformers in Inductive Reasoning Tasks
- 2025From Condensation to Rank Collapse: A Two-Stage Analysis of Transformer Training Dynamics
- 2025A Review of Developmental Interpretability in Large Language Models
- 2025Transformer Dynamics: A neuroscientific approach to interpretability of large language models
- 2026Understanding Large Language Models
- 2026LLMs Explain't: A Post-Mortem on Semantic Interpretability in Transformer Models
- 2026A Single Layer to Explain Them All:Understanding Massive Activations in Large Language Models