IAAIHierarchical Reasoning Model
Hierarchical Reasoning Model
A brain-inspired recurrent architecture with two interdependent modules: a slow high-level module for abstract planning and a fast low-level module for detailed computation, solving sequential reasoning tasks in one forward pass without intermediate supervision.
Inside the method
[ L-module × T → H-module update ] × N
Simplified conceptual schematic. Consult the paper for the complete architecture.
- Recurrence family
- Core recurrence
- Depth control
- Nested two-timescale recurrence
- KV / state strategy
- Not applicable (puzzle solver)
Reading note
With 27M parameters and about 1000 training examples, reports near-perfect results on hard Sudoku and large mazes and strong ARC performance without pretraining or CoT data. Later analyses (for example, Dissecting HRM and TRM) question how much the hierarchy itself contributes.
Sources checked 2026-10-28. This catalog does not imply independent reproduction.
Related works
Cite this work
@misc{hrm2025,
title = {Hierarchical Reasoning Model},
author = {Guan Wang and Jin Li and Yuhao Sun and Xing Chen and Changling Liu and Yue Wu and Meng Lu and Sen Song and Yasin Abbasi Yadkori},
year = {2025},
eprint = {2506.21734},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2506.21734}
}