Smallthinker: A Family Of Efficient Large Language Models Natively Trained For Local Deployment | Awesome LLM Papers Contribute to Awesome LLM Papers

Awesome LLM Papers

A curated collection of research papers on Large Language Models (LLMs), transformers, prompting, fine-tuning, and multi-modal systems. Maintained by Sean Moran.

Contact Sean Moran about this survey or website. Made with Jekyll and Hyde.

Smallthinker: A Family Of Efficient Large Language Models Natively Trained For Local Deployment

Yixin Song, Zhenliang Xue, Dongliang Wei, Feiyang Chen, Jianxiang Gao, Junchen Liu, Hangyu Liang, Guangshuo Qin, Chengrong Tian, Bo Wen, Longyu Zhao, Xinrui Zheng, Zeyu Mi, Haibo Chen . No Venue 2025

[Paper] [Paper]   Search on Google Scholar   Search on Semantic Scholar
Efficiency Memory & Context Model Architecture

While frontier large language models (LLMs) continue to push capability boundaries, their deployment remains confined to GPU-powered cloud infrastructure. We challenge this paradigm with SmallThinker, a family of LLMs natively designed - not adapted - for the unique constraints of local devices: weak computational power, limited memory, and slow storage. Unlike traditional approaches that mainly compress existing models built for clouds, we architect SmallThinker from the ground up to thrive within these limitations. Our innovation lies in a deployment-aware architecture that transforms constraints into design principles. First, We introduce a two-level sparse structure combining fine-grained Mixture-of-Experts (MoE) with sparse feed-forward networks, drastically reducing computational demands without sacrificing model capacity. Second, to conquer the I/O bottleneck of slow storage, we design a pre-attention router that enables our co-designed inference engine to prefetch expert parameters from storage while computing attention, effectively hiding storage latency that would otherwise cripple on-device inference. Third, for memory efficiency, we utilize NoPE-RoPE hybrid sparse attention mechanism to slash KV cache requirements. We release SmallThinker-4B-A0.6B and SmallThinker-21B-A3B, which achieve state-of-the-art performance scores and even outperform larger LLMs. Remarkably, our co-designed system mostly eliminates the need for expensive GPU hardware: with Q4_0 quantization, both models exceed 20 tokens/s on ordinary consumer CPUs, while consuming only 1GB and 8GB of memory respectively. SmallThinker is publicly available at hf.co/PowerInfer/SmallThinker-4BA0.6B-Instruct and hf.co/PowerInfer/SmallThinker-21BA3B-Instruct.

https://huggingface.co/discussions/paper/68883e4aaf872d625c10c64d

Similar Work