Pre-train and Fine-tune: Recommenders as Large Models

2025-01-24

AI Summary

To address the dual challenges of dynamic user interest evolution and prohibitively high fine-tuning costs for large language models (LLMs) in recommender systems, this paper proposes a two-stage fine-tuning paradigm grounded in information bottleneck theory: knowledge compression followed by knowledge matching. We introduce Information-Aware Adaptive Kernel (IAK) fine-tuning—the first technique explicitly designed for recommendation—enabling interpretable and generalizable lightweight adaptation. The method balances theoretical rigor with engineering practicality, supporting billion-scale online inference deployment. Evaluated on the homepage of a leading food-delivery platform, it achieves a 2.3% offline AUC gain and statistically significant improvements in online click-through rate (CTR) and gross merchandise volume (GMV). Our framework establishes a reusable, production-ready engineering paradigm for large-scale recommendation fine-tuning.

Technology Category

• Data Mining & Knowledge Management: Recommender Systems
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

In reality, users have different interests in different periods, regions, scenes, etc. Such changes in interest are so drastic that they are difficult to be captured by recommenders. Existing multi-domain learning can alleviate this problem. However, the structure of the industrial recommendation system is complex, the amount of data is huge, and the training cost is extremely high, so it is difficult to modify the structure of the industrial recommender and re-train it. To fill this gap, we consider recommenders as large pre-trained models and fine-tune them. We first propose the theory of the information bottleneck for fine-tuning and present an explanation for the fine-tuning technique in recommenders. To tailor for recommendation, we design an information-aware adaptive kernel (IAK) technique to fine-tune the pre-trained recommender. Specifically, we define fine-tuning as two phases: knowledge compression and knowledge matching and let the training stage of IAK explicitly approximate these two phases. Our proposed approach designed from the essence of fine-tuning is well interpretable. Extensive online and offline experiments show the superiority of our proposed method. Besides, we also share unique and important lessons we learned when deploying the method in a large-scale online platform. We also present the potential issues of fine-tuning techniques in recommendation systems and the corresponding solutions. The recommender with IAK technique has been deployed on the homepage of a billion-scale online food platform for several months and has yielded considerable profits in our business.

Problem

• Temporal-Spatial Context
• Recommendation Accuracy
• Model Fine-tuning Efficiency

Innovation

• IAK
• Knowledge Distillation and Matching
• Adaptive Recommendation Systems