Data Efficiency for Large Recommendation Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jain, Kshitij, Xie, Jingru, Regan, Kevin, Chen, Cheng, Han, Jie, Li, Steve, Li, Zhuoshu, Phillips, Todd, Sussman, Myles, Troup, Matt, Yu, Angel, Zhuo, Jia
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916453628772352
author Jain, Kshitij
Xie, Jingru
Regan, Kevin
Chen, Cheng
Han, Jie
Li, Steve
Li, Zhuoshu
Phillips, Todd
Sussman, Myles
Troup, Matt
Yu, Angel
Zhuo, Jia
author_facet Jain, Kshitij
Xie, Jingru
Regan, Kevin
Chen, Cheng
Han, Jie
Li, Steve
Li, Zhuoshu
Phillips, Todd
Sussman, Myles
Troup, Matt
Yu, Angel
Zhuo, Jia
contents Large recommendation models (LRMs) are fundamental to the multi-billion dollar online advertising industry, processing massive datasets of hundreds of billions of examples before transitioning to continuous online training to adapt to rapidly changing user behavior. The massive scale of data directly impacts both computational costs and the speed at which new methods can be evaluated (R&D velocity). This paper presents actionable principles and high-level frameworks to guide practitioners in optimizing training data requirements. These strategies have been successfully deployed in Google's largest Ads CTR prediction models and are broadly applicable beyond LRMs. We outline the concept of data convergence, describe methods to accelerate this convergence, and finally, detail how to optimally balance training data volume with model size.
format Preprint
id arxiv_https___arxiv_org_abs_2410_18111
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Data Efficiency for Large Recommendation Models
Jain, Kshitij
Xie, Jingru
Regan, Kevin
Chen, Cheng
Han, Jie
Li, Steve
Li, Zhuoshu
Phillips, Todd
Sussman, Myles
Troup, Matt
Yu, Angel
Zhuo, Jia
Information Retrieval
Machine Learning
Large recommendation models (LRMs) are fundamental to the multi-billion dollar online advertising industry, processing massive datasets of hundreds of billions of examples before transitioning to continuous online training to adapt to rapidly changing user behavior. The massive scale of data directly impacts both computational costs and the speed at which new methods can be evaluated (R&D velocity). This paper presents actionable principles and high-level frameworks to guide practitioners in optimizing training data requirements. These strategies have been successfully deployed in Google's largest Ads CTR prediction models and are broadly applicable beyond LRMs. We outline the concept of data convergence, describe methods to accelerate this convergence, and finally, detail how to optimally balance training data volume with model size.
title Data Efficiency for Large Recommendation Models
topic Information Retrieval
Machine Learning
url https://arxiv.org/abs/2410.18111