Hierarchical Reasoning Model
This addresses the problem of inefficient and unstable reasoning in AI systems, offering a potential transformative advancement for general-purpose reasoning, though it appears incremental relative to existing hierarchical or recurrent approaches.
The paper tackles the challenge of brittle and data-intensive reasoning in AI by proposing the Hierarchical Reasoning Model (HRM), which achieves nearly perfect performance on complex tasks like Sudoku and maze path-finding with only 27 million parameters and 1000 training samples, outperforming larger models on benchmarks such as the Abstraction and Reasoning Corpus.
Reasoning, the process of devising and executing complex goal-oriented action sequences, remains a critical challenge in AI. Current large language models (LLMs) primarily employ Chain-of-Thought (CoT) techniques, which suffer from brittle task decomposition, extensive data requirements, and high latency. Inspired by the hierarchical and multi-timescale processing in the human brain, we propose the Hierarchical Reasoning Model (HRM), a novel recurrent architecture that attains significant computational depth while maintaining both training stability and efficiency. HRM executes sequential reasoning tasks in a single forward pass without explicit supervision of the intermediate process, through two interdependent recurrent modules: a high-level module responsible for slow, abstract planning, and a low-level module handling rapid, detailed computations. With only 27 million parameters, HRM achieves exceptional performance on complex reasoning tasks using only 1000 training samples. The model operates without pre-training or CoT data, yet achieves nearly perfect performance on challenging tasks including complex Sudoku puzzles and optimal path finding in large mazes. Furthermore, HRM outperforms much larger models with significantly longer context windows on the Abstraction and Reasoning Corpus (ARC), a key benchmark for measuring artificial general intelligence capabilities. These results underscore HRM's potential as a transformative advancement toward universal computation and general-purpose reasoning systems.