MINAR: Mechanistic Interpretability for Neural Algorithmic Reasoning
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866918355716276224 |
|---|---|
| author | He, Jesse Jenne, Helen Vargas, Max Brown, Davis Mishne, Gal Wang, Yusu Kvinge, Henry |
| author_facet | He, Jesse Jenne, Helen Vargas, Max Brown, Davis Mishne, Gal Wang, Yusu Kvinge, Henry |
| contents | The recent field of neural algorithmic reasoning (NAR) studies the ability of graph neural networks (GNNs) to emulate classical algorithms like Bellman-Ford, a phenomenon known as algorithmic alignment. At the same time, recent advances in large language models (LLMs) have spawned the study of mechanistic interpretability, which aims to identify granular model components like circuits that perform specific computations. In this work, we introduce Mechanistic Interpretability for Neural Algorithmic Reasoning (MINAR), an efficient circuit discovery toolbox that adapts attribution patching methods from mechanistic interpretability to the GNN setting. We show through two case studies that MINAR recovers faithful neuron-level circuits from GNNs trained on algorithmic tasks. Our study sheds new light on the process of circuit formation and pruning during training, as well as giving new insight into how GNNs trained to perform multiple tasks in parallel reuse circuit components for related tasks. Our code is available at https://github.com/pnnl/MINAR. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_21442 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MINAR: Mechanistic Interpretability for Neural Algorithmic Reasoning He, Jesse Jenne, Helen Vargas, Max Brown, Davis Mishne, Gal Wang, Yusu Kvinge, Henry Machine Learning Artificial Intelligence The recent field of neural algorithmic reasoning (NAR) studies the ability of graph neural networks (GNNs) to emulate classical algorithms like Bellman-Ford, a phenomenon known as algorithmic alignment. At the same time, recent advances in large language models (LLMs) have spawned the study of mechanistic interpretability, which aims to identify granular model components like circuits that perform specific computations. In this work, we introduce Mechanistic Interpretability for Neural Algorithmic Reasoning (MINAR), an efficient circuit discovery toolbox that adapts attribution patching methods from mechanistic interpretability to the GNN setting. We show through two case studies that MINAR recovers faithful neuron-level circuits from GNNs trained on algorithmic tasks. Our study sheds new light on the process of circuit formation and pruning during training, as well as giving new insight into how GNNs trained to perform multiple tasks in parallel reuse circuit components for related tasks. Our code is available at https://github.com/pnnl/MINAR. |
| title | MINAR: Mechanistic Interpretability for Neural Algorithmic Reasoning |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2602.21442 |