Pre-trained LLMs Meet Sequential Recommenders: Efficient User-Centric Knowledge Distillation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Severin, Nikita, Kartushov, Danil, Urzhumov, Vladislav, Kulikov, Vladislav, Konovalova, Oksana, Grishanov, Alexey, Klenitskiy, Anton, Fatkulin, Artem, Vasilev, Alexey, Savchenko, Andrey, Makarov, Ilya
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908989445373952
author Severin, Nikita
Kartushov, Danil
Urzhumov, Vladislav
Kulikov, Vladislav
Konovalova, Oksana
Grishanov, Alexey
Klenitskiy, Anton
Fatkulin, Artem
Vasilev, Alexey
Savchenko, Andrey
Makarov, Ilya
author_facet Severin, Nikita
Kartushov, Danil
Urzhumov, Vladislav
Kulikov, Vladislav
Konovalova, Oksana
Grishanov, Alexey
Klenitskiy, Anton
Fatkulin, Artem
Vasilev, Alexey
Savchenko, Andrey
Makarov, Ilya
contents Sequential recommender systems have achieved significant success in modeling temporal user behavior but remain limited in capturing rich user semantics beyond interaction patterns. Large Language Models (LLMs) present opportunities to enhance user understanding with their reasoning capabilities, yet existing integration approaches create prohibitive inference costs in real time. To address these limitations, we present a novel knowledge distillation method that utilizes textual user profile generated by pre-trained LLMs into sequential recommenders without requiring LLM inference at serving time. The resulting approach maintains the inference efficiency of traditional sequential models while requiring neither architectural modifications nor LLM fine-tuning.
format Preprint
id arxiv_https___arxiv_org_abs_2604_21536
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Pre-trained LLMs Meet Sequential Recommenders: Efficient User-Centric Knowledge Distillation
Severin, Nikita
Kartushov, Danil
Urzhumov, Vladislav
Kulikov, Vladislav
Konovalova, Oksana
Grishanov, Alexey
Klenitskiy, Anton
Fatkulin, Artem
Vasilev, Alexey
Savchenko, Andrey
Makarov, Ilya
Information Retrieval
Artificial Intelligence
H.3.3; I.2.7
Sequential recommender systems have achieved significant success in modeling temporal user behavior but remain limited in capturing rich user semantics beyond interaction patterns. Large Language Models (LLMs) present opportunities to enhance user understanding with their reasoning capabilities, yet existing integration approaches create prohibitive inference costs in real time. To address these limitations, we present a novel knowledge distillation method that utilizes textual user profile generated by pre-trained LLMs into sequential recommenders without requiring LLM inference at serving time. The resulting approach maintains the inference efficiency of traditional sequential models while requiring neither architectural modifications nor LLM fine-tuning.
title Pre-trained LLMs Meet Sequential Recommenders: Efficient User-Centric Knowledge Distillation
topic Information Retrieval
Artificial Intelligence
H.3.3; I.2.7
url https://arxiv.org/abs/2604.21536