From Features to Transformers: Redefining Ranking for Scalable Impact

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Borisyuk, Fedor, Hertel, Lars, Parameswaran, Ganesh, Srivastava, Gaurav, Ramanujam, Sudarshan Srinivasa, Ocejo, Borja, Du, Peng, Akterskii, Andrei, Daftary, Neil, Tang, Shao, Sun, Daqi, Xiao, Qiang Charles, Nathani, Deepesh, Kothari, Mohit, Dai, Yun, Li, Guoyao, Gupta, Aman
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914312719695872
author Borisyuk, Fedor
Hertel, Lars
Parameswaran, Ganesh
Srivastava, Gaurav
Ramanujam, Sudarshan Srinivasa
Ocejo, Borja
Du, Peng
Akterskii, Andrei
Daftary, Neil
Tang, Shao
Sun, Daqi
Xiao, Qiang Charles
Nathani, Deepesh
Kothari, Mohit
Dai, Yun
Li, Guoyao
Gupta, Aman
author_facet Borisyuk, Fedor
Hertel, Lars
Parameswaran, Ganesh
Srivastava, Gaurav
Ramanujam, Sudarshan Srinivasa
Ocejo, Borja
Du, Peng
Akterskii, Andrei
Daftary, Neil
Tang, Shao
Sun, Daqi
Xiao, Qiang Charles
Nathani, Deepesh
Kothari, Mohit
Dai, Yun
Li, Guoyao
Gupta, Aman
contents We present LiGR, a large-scale ranking framework developed at LinkedIn that brings state-of-the-art transformer-based modeling architectures into production. We introduce a modified transformer architecture that incorporates learned normalization and simultaneous set-wise attention to user history and ranked items. This architecture enables several breakthrough achievements, including: (1) the deprecation of most manually designed feature engineering, outperforming the prior state-of-the-art system using only few features (compared to hundreds in the baseline), (2) validation of the scaling law for ranking systems, showing improved performance with larger models, more training data, and longer context sequences, and (3) simultaneous joint scoring of items in a set-wise manner, leading to automated improvements in diversity. To enable efficient serving of large ranking models, we describe techniques to scale inference effectively using single-pass processing of user history and set-wise attention. We also summarize key insights from various ablation studies and A/B tests, highlighting the most impactful technical approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2502_03417
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Features to Transformers: Redefining Ranking for Scalable Impact
Borisyuk, Fedor
Hertel, Lars
Parameswaran, Ganesh
Srivastava, Gaurav
Ramanujam, Sudarshan Srinivasa
Ocejo, Borja
Du, Peng
Akterskii, Andrei
Daftary, Neil
Tang, Shao
Sun, Daqi
Xiao, Qiang Charles
Nathani, Deepesh
Kothari, Mohit
Dai, Yun
Li, Guoyao
Gupta, Aman
Machine Learning
We present LiGR, a large-scale ranking framework developed at LinkedIn that brings state-of-the-art transformer-based modeling architectures into production. We introduce a modified transformer architecture that incorporates learned normalization and simultaneous set-wise attention to user history and ranked items. This architecture enables several breakthrough achievements, including: (1) the deprecation of most manually designed feature engineering, outperforming the prior state-of-the-art system using only few features (compared to hundreds in the baseline), (2) validation of the scaling law for ranking systems, showing improved performance with larger models, more training data, and longer context sequences, and (3) simultaneous joint scoring of items in a set-wise manner, leading to automated improvements in diversity. To enable efficient serving of large ranking models, we describe techniques to scale inference effectively using single-pass processing of user history and set-wise attention. We also summarize key insights from various ablation studies and A/B tests, highlighting the most impactful technical approaches.
title From Features to Transformers: Redefining Ranking for Scalable Impact
topic Machine Learning
url https://arxiv.org/abs/2502.03417