Hugging Face Trending Papers

RecoGEM: A Serving System for Production-Scale Generative Recommendation

RecoGEM is a production serving system designed for generative recommendation (GR) models, which shift from scoring to autoregressively generating item identifiers. It decouples runtime responsibilities from the execution engine, optimizing self‑attention, candidate selection, and execution planning for GR’s short token sequences and large candidate pools. Deployed for nine months, RecoGEM serves hundreds of millions of users, supports over ten GR model families, and achieves more than 1,000 QPS per instance with P99 latency below 35 ms, improving throughput by 2.88× over baselines.

arXiv AI
Sep 25

DeGRe: Dense-supervised Generative Reranking for Recommendation

DeGRe is a dense‑supervised generative reranking framework designed to improve multi‑stage recommender systems by addressing label bias and credit assignment issues. It uses an offline Lookahead Evaluator with beam search to generate dense supervision signals, which are distilled into a lightweight Online Generator that can perform efficient greedy decoding at inference time. Experiments show that DeGRe outperforms baselines on public benchmarks and industrial datasets, and it has been successfully deployed on Taobao Flash Shopping to enhance online recommendations.

By Chaotian Song, Jingyao Zhang, Chenghao Chen, Zisen Sang, Dehai Zhao, Guodong Cao, Boxi Wu, Deng Cai, Jia Jia
arXiv AI
Jul 1

GR2 Technical Report

arXiv:2606. 31984v1 Announce Type: cross Abstract: Industrial recommendation systems serve billions of users through a multi-stage funnel -- retrieval, early-stage ranking, and re-ranking -- where the final re-ranking step disproportionately shapes user engagement and downstream performance, particularly for carousel and grid display formats.

By Yufei Li (Yongkang), Zaiwei Zhang (Yongkang), Mingfu Liang (Yongkang), Kavosh Asadi (Yongkang), Jay Xu (Yongkang), Jimmy Kim (Yongkang), Chongyang Bai (Yongkang), Jieyi Zhang (Yongkang), Hongye Xie (Yongkang), Prachi Agrawal (Yongkang), Dian Yu (Yongkang), Tianyi Chen (Yongkang), Jean-Pascal Billaud (Yongkang), Garret Buell (Yongkang), YK (Yongkang), Zhu (Yang), Sachin Patil (Yang), Brooke Bian (Yang), Zhou Fang (Yang), Kevin Huang (Yang), Shiva Sudanagunta (Yang), Yuzhen Huang (Yang), Emma Lu (Yang), Chris O'Brien (Yang), Yang Song (Yang), Lihong Li (Yang), Jacob Tao (Yang), Zhicheng Zhu (Yang), Chao Li (Yang), Gaoxiang Liu (Yang), Neil Wu (Yang), Zhongyin Hu (Yang), Li Han (Yang), Loki Chen (Yang), Ming Lei (Yang), Greg Rehm (Yang), Siyuan Song (Yang), Tianwei Zhang (Yang), Li Li (Yang), Ketan Singh (Yang), Yavuz Yetim (Yang), Ilyas Atishev (Yang), Satendra Gera (Yang), Ashkan Sadeghi (Yang), Rachel Yan (Yang), Nikko Mizutani (Yang), Shuaiwen Wang (Yang), Song Yang (Yang), Zhijing Li (Yang), Jiang Liu (Yang), Mengying Sun (Yang), Fei Tian (Yang), Xiaohan Wei (Yang), Chonglin Sun (Yang), Parish Aggarwal (Yang), Kaushik Rangadurai (Yang), Zhi Hua (Yang), Frank Shyu (Yang), Ruchit Sharma (Yang), Liyuan Li (Yang), Shike Mei (Yang), Wenlin Chen (Yang), Santanu Kolay (Yang), Ben Schulte (Yang), Deepak Chandra (Yang), Adam (Yang), Song, Sandeep Pandey, Xi Liu, Hamed Firooz, Luke Simon
arXiv AI
Sep 10

SequenceO1: End-to-End Ultra-Long (100K) Sequence Modeling in Recommendation with Low-Rank Caching

SequenceO1 is an end‑to‑end framework that enables ultra‑long (up to 100K interactions) sequence modeling for recommendation systems. It compresses raw user histories into a fixed‑size sketch using Sketch Attention and then models short‑term and long‑term interests with Target‑to‑History Cross Attention. The system incorporates low‑rank caching, batching, pipeline lift, and a FlashSA kernel to keep training and inference efficient, achieving consistent offline and online performance gains when deployed at full traffic on Douyin.

By Lin Guan, Jia-Qi Yang, Zhishan Zhao, Jiaqi Huang, Hangyu Wang, Longbin Li, Beichuan Zhang, Haonan Jiang, Jinan Ni, Xiangyu Fan, Xiaowen Li, Ziyao Ren, Yuhang Qi, Xiaolong Zhu, Xuanyuan Luo, Qiwei Chen, Yi Cheng, Lele Yu
arXiv Computation and Language
Sep 11

Auto-RecSys: Harnessing Autonomous Research Agents for Industry-Scale Recommender System

Auto-RecSys is an autonomous research system designed to scale long-horizon experimentation for industry‑scale recommendation models. It tackles long feedback loops and system complexity by enabling distributed asynchronous execution, centralized cross‑server memory, and a cognitive‑procedural separation that combines natural‑language skill files with deterministic scripts. The system incorporates a dual‑loop self‑evolving architecture—Execution Evolution and Idea Evolution loops—to refine operational playbooks and guide future experiments, thereby reducing human effort per cycle and improving reliability as playbooks mature.

By Ming Li, Dai Li, Xuying Ning, Bo Sun, Rui Li, Yi Zhang, Silvia Gong, Xuan Cao, Rui Li, Cornelia Carapcea, Qunshu Zhang, Zhigang Wang, Yinglong Xia, Andy Wang
arXiv AI
Aug 24

Difficulty-Aware Semantic-ID Optimization for Generative Recommendation

The paper introduces Difficulty‑Aware Semantic‑ID Optimization (DASO), a post‑training method for generative recommendation that improves tree‑structured item ranking. DASO profiles rollout groups by prefix‑match depth, reallocates a portion of candidates to prefix‑guided completions, and uses a SID‑prefix reward with an auxiliary SFT anchor to address target‑missing failures. On public benchmarks, DASO outperforms MiniOneRec‑style GRPO on 11 of 12 metrics and achieves the best results on 9 of 12 metrics, also improving level‑wise recall on an internal recommendation task.

By Xin Yu, Stephen Li, Sina Aghaei, Zifan Zhu, Jiamu Bai, Guanjie Huang, Bo Peng, Yiyao Liu, Lingzhou Xue
arXiv Machine Learning
Sep 23

Lightweight Ranking Heads: Accelerating Multi-Task Experimentation in Production Recommender Systems

The paper introduces Lightweight Ranking Heads (Light Heads), a framework that allows new prediction tasks to be added to large multi-task recommender systems without retraining the backbone model. By using stop‑gradients and stateless daily training, Light Heads isolate new tasks, preventing conflicts with existing ones. Deployed at YouTube scale, the approach cuts multi‑task experimentation cycles from weeks to days, enabling faster A/B testing and deployment of new ranking tasks.

By Sanjay Surendranath Girija, Aniruddh Nath, Li Wei, Yanhao Jiang, Shawn Andrews, Lukasz Heldt, Yi Wu, Aditya Mahajan, Mohit Sharma
arXiv Computation and Language
Sep 3

CORAL: An LLM-Native Harness for Production Recommender Systems

CORAL is an LLM‑native harness that automates continual optimization of production recommender systems. It operates in a closed loop: an agent observes system signals, reasons over past decisions, and uses tools—including a numerical optimizer—to reconfigure the recommender while staying within a fixed operating budget. In A/B experiments on two large social platforms, CORAL improved engagement without extra serving cost on one platform and reduced serving cost without harming engagement on the other, demonstrating that a single agentic loop can replace manual engineering for ongoing system tuning.

By Muhammad Rafay Azhar, Yuhang Zhou, Gilbert Jiang, Yuchen Wang, Rahul Sharma, Matthew DeSousa, Jiayi Liu, Xin Guo, Lizhu Zhang, Xiangjun Fan
arXiv Machine Learning
Aug 5

Target-Aware Early Stage Ranking

arXiv:2511. 21095v2 Announce Type: replace Abstract: Early Stage Ranking (ESR) in large-scale recommendation systems is dominated by ''user--item decoupling'' Two Tower architectures, which scale efficiently but cannot capture fine-grained, target-aware user--item interactions directly.

By Juhee Hong, Meng Liu, Shengzhi Wang, Jin Zhou, Xiaoheng Mao, Zhao Zhu, Ruochen Liu, Huihui Cheng, Leon Gao, Christopher Leung, Chandra Mouli Sekar, Yijia Liu, Boyang Yu, Tuan Trieu, Dawei Sun, Jeet Kanjani, Rui Li, Jing Qian, Xuan Cao, Minjie Fan, Mingze Gao
arXiv AI
Sep 15

Safety as a Constraint: Fine-Tuning a LLM Recommender to Explain Itself

The paper presents a method for fine‑tuning a large language model (LLM) recommender to generate personalized, non‑harmful explanations for its recommendations. By training two LLM‑judge reward models and using constrained GRPO, the authors achieve a significant increase in the PASS rate for all three criteria, from 0.649 to 0.956 on their own judges and from 0.677 to 0.931 on an independent judge. The fine‑tuned model maintains its original recommendation performance, demonstrating that LLM‑based recommenders can be adapted to complex tasks without loss of effectiveness.

By Jiashu He, Emma Yanyang Kong, JJ Tan, David Fagnan