Research
Causal Concept Graphs in LLM Latent Space for Stepwise Reasoning
arXiv:2603.10377v2 Announce Type: replace-cross Abstract: Sparse autoencoders can localize where concepts live in language models, but not how they interact during multi-step reasoning. We propose Cau
arXiv:2603.10377v2 Announce Type: replace-cross Abstract: Sparse autoencoders can localize where concepts live in language models, but not how they interact during multi-step reasoning. We propose Causal Concept Graphs (CCG): a directed acyclic graph over sparse, interpretable latent features, where edges capture learned causal dependencies between concepts. We combine task-conditioned sparse autoencoders for concept discovery with DAGMA-style differentiable structure learning for graph recovery and introduce the Causal Fidelity Score (CFS) to evaluate whether graph-guided interventions induce larger downstream effects than random ones. On ARC-Challenge, StrategyQA, and LogiQA with GPT-2 Medium, across five seeds (n{=}15 paired runs), CCG achieves CFS=5.654pm0.625, outperforming ROME-style tracing (3.382pm0.233), SAE-only ranking (2.479pm0.196), and a random baseline (1.032pm0.034), with p<0.0001 after Bonferroni correction. Learned graphs are sparse (5-6% edge density), domain-specific, and stable across seeds.
Related
- Mechanistic Interpretability of Antibody Language Models Using SAEs
- Schoenfeld's Anatomy of Mathematical Reasoning by Language Models
- A Survey of Inductive Reasoning for Large Language Models
- The Shape of Adversarial Influence: Characterizing LLM Latent Spaces with Persistent Homology
Source: arXiv cs.AI | 2026-04-27