OneRec Technical Report
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909790998888448 |
|---|---|
| author | Zhou, Guorui Deng, Jiaxin Zhang, Jinghao Cai, Kuo Ren, Lejian Luo, Qiang Wang, Qianqian Hu, Qigen Huang, Rui Wang, Shiyao Ding, Weifeng Li, Wuchao Luo, Xinchen Wang, Xingmei Cheng, Zexuan Zhang, Zixing Zhang, Bin Wang, Boxuan Ma, Chaoyi Song, Chengru Wang, Chenhui Wang, Di Meng, Dongxue Yang, Fan Zhang, Fangyu Jiang, Feng Zhang, Fuxing Wang, Gang Zhang, Guowang Li, Han Hu, Hengrui Lin, Hezheng Cheng, Hongtao Cao, Hongyang Wang, Huanjie Huang, Jiaming Chen, Jiapeng Liu, Jiaqiang Jia, Jinghui Gai, Kun Hu, Lantao Zeng, Liang Yu, Liao Wang, Qiang Zhou, Qidong Wang, Shengzhe He, Shihui Yang, Shuang Yang, Shujie Huang, Sui Wu, Tao He, Tiantian Gao, Tingting Yuan, Wei Liang, Xiao Xu, Xiaoxiao Liu, Xugang Wang, Yan Wang, Yi Liu, Yiwu Song, Yue Zhang, Yufei Wu, Yunfan Zhao, Yunfeng Liu, Zhanyu |
| author_facet | Zhou, Guorui Deng, Jiaxin Zhang, Jinghao Cai, Kuo Ren, Lejian Luo, Qiang Wang, Qianqian Hu, Qigen Huang, Rui Wang, Shiyao Ding, Weifeng Li, Wuchao Luo, Xinchen Wang, Xingmei Cheng, Zexuan Zhang, Zixing Zhang, Bin Wang, Boxuan Ma, Chaoyi Song, Chengru Wang, Chenhui Wang, Di Meng, Dongxue Yang, Fan Zhang, Fangyu Jiang, Feng Zhang, Fuxing Wang, Gang Zhang, Guowang Li, Han Hu, Hengrui Lin, Hezheng Cheng, Hongtao Cao, Hongyang Wang, Huanjie Huang, Jiaming Chen, Jiapeng Liu, Jiaqiang Jia, Jinghui Gai, Kun Hu, Lantao Zeng, Liang Yu, Liao Wang, Qiang Zhou, Qidong Wang, Shengzhe He, Shihui Yang, Shuang Yang, Shujie Huang, Sui Wu, Tao He, Tiantian Gao, Tingting Yuan, Wei Liang, Xiao Xu, Xiaoxiao Liu, Xugang Wang, Yan Wang, Yi Liu, Yiwu Song, Yue Zhang, Yufei Wu, Yunfan Zhao, Yunfeng Liu, Zhanyu |
| contents | Recommender systems have been widely used in various large-scale user-oriented platforms for many years. However, compared to the rapid developments in the AI community, recommendation systems have not achieved a breakthrough in recent years. For instance, they still rely on a multi-stage cascaded architecture rather than an end-to-end approach, leading to computational fragmentation and optimization inconsistencies, and hindering the effective application of key breakthrough technologies from the AI community in recommendation scenarios.
To address these issues, we propose OneRec, which reshapes the recommendation system through an end-to-end generative approach and achieves promising results. Firstly, we have enhanced the computational FLOPs of the current recommendation model by 10 $\times$ and have identified the scaling laws for recommendations within certain boundaries. Secondly, reinforcement learning techniques, previously difficult to apply for optimizing recommendations, show significant potential in this framework. Lastly, through infrastructure optimizations, we have achieved 23.7% and 28.8% Model FLOPs Utilization (MFU) on flagship GPUs during training and inference, respectively, aligning closely with the LLM community. This architecture significantly reduces communication and storage overhead, resulting in operating expense that is only 10.6% of traditional recommendation pipelines. Deployed in Kuaishou/Kuaishou Lite APP, it handles 25% of total queries per second, enhancing overall App Stay Time by 0.54% and 1.24%, respectively. Additionally, we have observed significant increases in metrics such as 7-day Lifetime, which is a crucial indicator of recommendation experience. We also provide practical lessons and insights derived from developing, optimizing, and maintaining a production-scale recommendation system with significant real-world impact. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_13695 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | OneRec Technical Report Zhou, Guorui Deng, Jiaxin Zhang, Jinghao Cai, Kuo Ren, Lejian Luo, Qiang Wang, Qianqian Hu, Qigen Huang, Rui Wang, Shiyao Ding, Weifeng Li, Wuchao Luo, Xinchen Wang, Xingmei Cheng, Zexuan Zhang, Zixing Zhang, Bin Wang, Boxuan Ma, Chaoyi Song, Chengru Wang, Chenhui Wang, Di Meng, Dongxue Yang, Fan Zhang, Fangyu Jiang, Feng Zhang, Fuxing Wang, Gang Zhang, Guowang Li, Han Hu, Hengrui Lin, Hezheng Cheng, Hongtao Cao, Hongyang Wang, Huanjie Huang, Jiaming Chen, Jiapeng Liu, Jiaqiang Jia, Jinghui Gai, Kun Hu, Lantao Zeng, Liang Yu, Liao Wang, Qiang Zhou, Qidong Wang, Shengzhe He, Shihui Yang, Shuang Yang, Shujie Huang, Sui Wu, Tao He, Tiantian Gao, Tingting Yuan, Wei Liang, Xiao Xu, Xiaoxiao Liu, Xugang Wang, Yan Wang, Yi Liu, Yiwu Song, Yue Zhang, Yufei Wu, Yunfan Zhao, Yunfeng Liu, Zhanyu Information Retrieval Recommender systems have been widely used in various large-scale user-oriented platforms for many years. However, compared to the rapid developments in the AI community, recommendation systems have not achieved a breakthrough in recent years. For instance, they still rely on a multi-stage cascaded architecture rather than an end-to-end approach, leading to computational fragmentation and optimization inconsistencies, and hindering the effective application of key breakthrough technologies from the AI community in recommendation scenarios. To address these issues, we propose OneRec, which reshapes the recommendation system through an end-to-end generative approach and achieves promising results. Firstly, we have enhanced the computational FLOPs of the current recommendation model by 10 $\times$ and have identified the scaling laws for recommendations within certain boundaries. Secondly, reinforcement learning techniques, previously difficult to apply for optimizing recommendations, show significant potential in this framework. Lastly, through infrastructure optimizations, we have achieved 23.7% and 28.8% Model FLOPs Utilization (MFU) on flagship GPUs during training and inference, respectively, aligning closely with the LLM community. This architecture significantly reduces communication and storage overhead, resulting in operating expense that is only 10.6% of traditional recommendation pipelines. Deployed in Kuaishou/Kuaishou Lite APP, it handles 25% of total queries per second, enhancing overall App Stay Time by 0.54% and 1.24%, respectively. Additionally, we have observed significant increases in metrics such as 7-day Lifetime, which is a crucial indicator of recommendation experience. We also provide practical lessons and insights derived from developing, optimizing, and maintaining a production-scale recommendation system with significant real-world impact. |
| title | OneRec Technical Report |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2506.13695 |