MTGR: Industrial-Scale Generative Recommendation Framework in Meituan

TL;DR

MTGR is a scalable generative recommendation framework that retains traditional model features, accelerates training and inference, and has been successfully deployed at Meituan, significantly improving efficiency and performance.

Contribution

The paper introduces MTGR, a novel recommendation framework that preserves cross features, accelerates training and inference, and supports large-scale deployment in industrial settings.

Findings

Achieved 65x FLOPs reduction for inference compared to DLRM.

Supported models with 10 to 100 times the computational complexity of DLRM.

Successfully deployed on Meituan's platform, handling main traffic.

Abstract

Scaling law has been extensively validated in many domains such as natural language processing and computer vision. In the recommendation system, recent work has adopted generative recommendations to achieve scalability, but their generative approaches require abandoning the carefully constructed cross features of traditional recommendation models. We found that this approach significantly degrades model performance, and scaling up cannot compensate for it at all. In this paper, we propose MTGR (Meituan Generative Recommendation) to address this issue. MTGR is modeling based on the HSTU architecture and can retain the original deep learning recommendation model (DLRM) features, including cross features. Additionally, MTGR achieves training and inference acceleration through user-level compression to ensure efficient scaling. We also propose Group-Layer Normalization (GLN) to enhance the performance of encoding within different semantic spaces and the dynamic masking strategy to avoid information leakage. We further optimize the training frameworks, enabling support for our models with 10 to 100 times computational complexity compared to the DLRM, without significant cost increases. MTGR achieved 65x FLOPs for single-sample forward inference compared to the DLRM model, resulting in the largest gain in nearly two years both offline and online. This breakthrough was successfully deployed on Meituan, the world's largest food delivery platform, where it has been handling the main traffic.