<ul><li>Recent studies challenge the assumption that only large language models (LLMs) can achieve competitive reasoning performance.</li><li>Small language models (SLMs) are also shown to have strong reasoning capabilities and are favored for their efficiency and deployability.</li><li>A systematic study examines 72 SLMs from six model families across 14 reasoning benchmarks.</li><li>The findings suggest that SLMs with strong reasoning abilities can be developed through structured training or post-training compression as efficient alternatives to LLMs.</li></ul>

Towards Reasoning Ability of Small Language Models

Discover more