arXiv cs.AI· Qi Xu, Zhigang Hua, Zijie Lei, Yan Xie, Jizhe Zhang, Lynn Lu, Delun Du, Sha Hua, Ren Mao, Ke Pan, Haomin Yu, Sandeep Pandey, Luke Simon·· 2 小时前AI 评分39
MIS：一次推理实现高效整页排序的 prefix-aware 非自回归框架
Towards Efficient Whole Page Ranking with One-Pass Inference
AI 导读
研究者提出多物品打分框架 MIS，在非自回归式整页排序中引入序列上下文，兼顾质量与效率。在 KuaiRand-1K 和 RL4RS 上，MIS 接近自回归 CGA 基线的生成质量，同时 CPU 解码分别加速 4.07 倍和 3.54 倍。在大型社交视频平台的随机评估中，MIS 将 realization 指标提升 0.104%、观看时长提升 0.058%，排序服务 CPU 消耗降低 9.73%。
正文 · 原文
Authors:Qi Xu, Zhigang Hua, Zijie Lei, Yan Xie, Jizhe Zhang, Lynn Lu, Delun Du, Sha Hua, Ren Mao, Ke Pan, Haomin Yu, Sandeep Pandey, Luke Simon
Abstract:Whole-page ranking jointly optimizes advertisements and organic content to balance business outcomes and user engagement under complex layout and business constraints. Existing non-autoregressive (NAR) approaches enable efficient page generation but typically score candidate items independently of previously selected items, limiting their ability to capture order-dependent interactions. Autoregressive (AR) methods address this limitation through sequential conditioning, but incur substantial inference overhead that challenges large-scale deployment. Bridging this quality-efficiency gap remains a challenge in industrial whole-page ranking.
We propose Multi-Item Scoring (MIS), a prefix-aware NAR-style framework that incorporates sequential context into efficient multi-item scoring. MIS integrates global candidate interactions, localized prefix conditioning, and position-aware scoring within a shared representation architecture. By computing reusable representations once per request and performing lightweight prefix-conditioned scoring during constrained beam search, MIS captures important sequential dependencies without repeated full-model inference.
Experiments on KuaiRand-1K and RL4RS demonstrate that MIS approaches the generation quality of an autoregressive CGA baseline while achieving 4.07x and 3.54x faster CPU decoding, respectively. In a randomized evaluation on a large-scale social-video platform, MIS improves the realization metric by 0.104% and watch time by 0.058%, while reducing ranking-service CPU consumption by 9.73%. These results demonstrate that prefix-aware multi-item scoring can narrow the quality gap between NAR and AR generation while retaining the efficiency required for industrial scalability.
| Comments: | 12 pages, 2 figures, 5 tables |
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (stat.ML) |
| Cite as: | arXiv:2611.05542 [cs.AI] |
| (or arXiv:2611.05542v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.05542 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Qi Xu [view email]
[v1]
Tue, 3 Nov 2026 19:33:22 UTC (625 KB)
来源：arXiv cs.AI · arxiv.org