Large Scale Retrieval for the LinkedIn Feed using Causal Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908596742127616 |
|---|---|
| author | Ramanujam, Sudarshan Srinivasa Alonso, Antonio Kataria, Saurabh Dangi, Siddharth Gupta, Akhilesh Tiwana, Birjodh Singh Somaiya, Manas Simon, Luke Byrne, David Ha, Sojeong Zhou, Sen Akterskii, Andrei Liu, Zhanglong Sriram, Samira Xiong, Crescent Pei, Zhoutao Shao, Angela Li, Alex Xiao, Annie Kolb, Caitlin Kistler, Thomas Moore, Zach Firooz, Hamed |
| author_facet | Ramanujam, Sudarshan Srinivasa Alonso, Antonio Kataria, Saurabh Dangi, Siddharth Gupta, Akhilesh Tiwana, Birjodh Singh Somaiya, Manas Simon, Luke Byrne, David Ha, Sojeong Zhou, Sen Akterskii, Andrei Liu, Zhanglong Sriram, Samira Xiong, Crescent Pei, Zhoutao Shao, Angela Li, Alex Xiao, Annie Kolb, Caitlin Kistler, Thomas Moore, Zach Firooz, Hamed |
| contents | In large scale recommendation systems like the LinkedIn Feed, the retrieval stage is critical for narrowing hundreds of millions of potential candidates to a manageable subset for ranking. LinkedIn's Feed serves suggested content from outside of the member's network (based on the member's topical interests), where 2000 candidates are retrieved from a pool of hundreds of millions candidate with a latency budget of a few milliseconds and inbound QPS of several thousand per second. This paper presents a novel retrieval approach that fine-tunes a large causal language model (Meta's LLaMA 3) as a dual encoder to generate high quality embeddings for both users (members) and content (items), using only textual input. We describe the end to end pipeline, including prompt design for embedding generation, techniques for fine-tuning at LinkedIn's scale, and infrastructure for low latency, cost effective online serving. We share our findings on how quantizing numerical features in the prompt enables the information to get properly encoded in the embedding, facilitating greater alignment between the retrieval and ranking layer. The system was evaluated using offline metrics and an online A/B test, which showed substantial improvements in member engagement. We observed significant gains among newer members, who often lack strong network connections, indicating that high-quality suggested content aids retention. This work demonstrates how generative language models can be effectively adapted for real time, high throughput retrieval in industrial applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_14223 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Large Scale Retrieval for the LinkedIn Feed using Causal Language Models Ramanujam, Sudarshan Srinivasa Alonso, Antonio Kataria, Saurabh Dangi, Siddharth Gupta, Akhilesh Tiwana, Birjodh Singh Somaiya, Manas Simon, Luke Byrne, David Ha, Sojeong Zhou, Sen Akterskii, Andrei Liu, Zhanglong Sriram, Samira Xiong, Crescent Pei, Zhoutao Shao, Angela Li, Alex Xiao, Annie Kolb, Caitlin Kistler, Thomas Moore, Zach Firooz, Hamed Information Retrieval Artificial Intelligence In large scale recommendation systems like the LinkedIn Feed, the retrieval stage is critical for narrowing hundreds of millions of potential candidates to a manageable subset for ranking. LinkedIn's Feed serves suggested content from outside of the member's network (based on the member's topical interests), where 2000 candidates are retrieved from a pool of hundreds of millions candidate with a latency budget of a few milliseconds and inbound QPS of several thousand per second. This paper presents a novel retrieval approach that fine-tunes a large causal language model (Meta's LLaMA 3) as a dual encoder to generate high quality embeddings for both users (members) and content (items), using only textual input. We describe the end to end pipeline, including prompt design for embedding generation, techniques for fine-tuning at LinkedIn's scale, and infrastructure for low latency, cost effective online serving. We share our findings on how quantizing numerical features in the prompt enables the information to get properly encoded in the embedding, facilitating greater alignment between the retrieval and ranking layer. The system was evaluated using offline metrics and an online A/B test, which showed substantial improvements in member engagement. We observed significant gains among newer members, who often lack strong network connections, indicating that high-quality suggested content aids retention. This work demonstrates how generative language models can be effectively adapted for real time, high throughput retrieval in industrial applications. |
| title | Large Scale Retrieval for the LinkedIn Feed using Causal Language Models |
| topic | Information Retrieval Artificial Intelligence |
| url | https://arxiv.org/abs/2510.14223 |