Saved in:
Bibliographic Details
Main Authors: Behdin, Kayhan, Fatahibaarzi, Ata, Song, Qingquan, Dai, Yun, Gupta, Aman, Wang, Zhipeng, Tang, Shao, Sang, Hejian, Dexter, Gregory, Zhu, Sirou, Zhu, Siyu, Dharamsi, Tejas, Kothapalli, Vignesh, Fu, Zhoutong, Cao, Yihan, Hsu, Pin-Lun, Borisyuk, Fedor, Pillai, Natesh, Simon, Luke, Mazumder, Rahul
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.14305
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911232518258688
author Behdin, Kayhan
Fatahibaarzi, Ata
Song, Qingquan
Dai, Yun
Gupta, Aman
Wang, Zhipeng
Tang, Shao
Sang, Hejian
Dexter, Gregory
Zhu, Sirou
Zhu, Siyu
Dharamsi, Tejas
Kothapalli, Vignesh
Fu, Zhoutong
Cao, Yihan
Hsu, Pin-Lun
Borisyuk, Fedor
Pillai, Natesh
Simon, Luke
Mazumder, Rahul
author_facet Behdin, Kayhan
Fatahibaarzi, Ata
Song, Qingquan
Dai, Yun
Gupta, Aman
Wang, Zhipeng
Tang, Shao
Sang, Hejian
Dexter, Gregory
Zhu, Sirou
Zhu, Siyu
Dharamsi, Tejas
Kothapalli, Vignesh
Fu, Zhoutong
Cao, Yihan
Hsu, Pin-Lun
Borisyuk, Fedor
Pillai, Natesh
Simon, Luke
Mazumder, Rahul
contents Large language models (LLMs) have demonstrated remarkable performance across a wide range of industrial applications, from search and recommendation systems to generative tasks. Although scaling laws indicate that larger models generally yield better generalization and performance, their substantial computational requirements often render them impractical for many real-world scenarios at scale. In this paper, we present a comprehensive set of insights for training and deploying small language models (SLMs) that deliver high performance for a variety of industry use cases. We focus on two key techniques: (1) knowledge distillation and (2) model compression via structured pruning and quantization. These approaches enable SLMs to retain much of the quality of their larger counterparts while significantly reducing training/serving costs and latency. We detail the impact of these techniques on a variety of use cases in a large professional social network platform and share deployment lessons, including hardware optimization strategies that improve speed and throughput for both predictive and reasoning-based applications in Recommendation Systems.
format Preprint
id arxiv_https___arxiv_org_abs_2502_14305
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scaling Down, Serving Fast: Compressing and Deploying Efficient LLMs for Recommendation Systems
Behdin, Kayhan
Fatahibaarzi, Ata
Song, Qingquan
Dai, Yun
Gupta, Aman
Wang, Zhipeng
Tang, Shao
Sang, Hejian
Dexter, Gregory
Zhu, Sirou
Zhu, Siyu
Dharamsi, Tejas
Kothapalli, Vignesh
Fu, Zhoutong
Cao, Yihan
Hsu, Pin-Lun
Borisyuk, Fedor
Pillai, Natesh
Simon, Luke
Mazumder, Rahul
Information Retrieval
Machine Learning
Large language models (LLMs) have demonstrated remarkable performance across a wide range of industrial applications, from search and recommendation systems to generative tasks. Although scaling laws indicate that larger models generally yield better generalization and performance, their substantial computational requirements often render them impractical for many real-world scenarios at scale. In this paper, we present a comprehensive set of insights for training and deploying small language models (SLMs) that deliver high performance for a variety of industry use cases. We focus on two key techniques: (1) knowledge distillation and (2) model compression via structured pruning and quantization. These approaches enable SLMs to retain much of the quality of their larger counterparts while significantly reducing training/serving costs and latency. We detail the impact of these techniques on a variety of use cases in a large professional social network platform and share deployment lessons, including hardware optimization strategies that improve speed and throughput for both predictive and reasoning-based applications in Recommendation Systems.
title Scaling Down, Serving Fast: Compressing and Deploying Efficient LLMs for Recommendation Systems
topic Information Retrieval
Machine Learning
url https://arxiv.org/abs/2502.14305