# Next-Scale Generative Reranking: A Tree-based Generative Rerank Method at Meituan

Source: [Next-Scale Generative Reranking: A Tree-based Generative Rerank Method at Meituan](<https://www.sota2.com/research/paper/69d5e12f7d0172140ec75c1f>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-09T05:00:04.534Z

Authors: Shuli Wang, Changhao Li, Ke Fan, Senjie Kou Junwei Yin, Chi Wang, Yinhua Zhu, Haitao Wang, Xingxing Wang

Published: 2026-04-07T01:35:20.000Z

## Abstract

In modern multi-stage recommendation systems, reranking plays a critical role by modeling contextual information. Due to inherent challenges such as the combinatorial space complexity, an increasing number of methods adopt the generative paradigm: the generator produces the optimal list during inference, while an evaluator guides the generator's optimization during the training phase. However, these methods still face two problems. Firstly, these generators fail to produce optimal generation results due to the lack of both local and global perspectives, regardless of whether the generation strategy is autoregressive or non-autoregressive. Secondly, the goal inconsistency problem between the generator and the evaluator during training complicates the guidance signal and leading to suboptimal performance. To address these issues, we propose the \\textbf{N}ext-\\textbf{S}cale \\textbf{G}eneration \\textbf{R}eranking (NSGR), a tree-based generative framework. Specifically, we introduce a next-scale generator (NSG) that progressively expands a recommendation list from user interests in a coarse-to-fine manner, balancing global and local perspectives. Furthermore, we design a multi-scale neighbor loss, which leverages a tree-based multi-scale evaluator (MSE) to provide scale-specific guidance to the NSG at each scale. Extensive experiments on public and industrial datasets validate the effectiveness of NSGR. And NSGR has been successfully deployed on the Meituan food delivery platform.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Reranking on Meituan](<https://www.sota2.com/research/sota/reranking-on-meituan.md>): Reranking; Meituan; AUC: 89.02.

- [Reranking on Taobao Ad dataset](<https://www.sota2.com/research/sota/reranking-on-taobao-ad-dataset.md>): Reranking; Taobao Ad dataset; AUC: 63.96.

- [Reranking on Meituan (HR@1%, HR@10%)](<https://www.sota2.com/research/sota/reranking-on-meituan-hr-at-1-percent-hr-at-10-percent.md>): Reranking; Meituan; HR@1%: 86.1.

- [Reranking on Meituan food delivery platform online A/B test](<https://www.sota2.com/research/sota/reranking-on-meituan-food-delivery-platform-online-a-b-test.md>): Reranking; Meituan food delivery platform online A/B (test); CTR Uplift: 2.89.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.05314>)

[PDF](<https://arxiv.org/pdf/2604.05314.pdf>)
