Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Mosaic, a user modeling platform that utilizes a fleet of four diverse model families to enhance user embeddings for recommendation systems at Meta. By employing novel techniques such as Multi-task Relations Mining and Cosine Redundancy Loss, Mosaic effectively maximizes the information contribution of each specialist, leading to substantial improvements in both offline and online performance metrics. The results demonstrate that this specialized approach to user representation can significantly enhance retrieval, ranking, and integrity tasks across the platform.
Mosaic's fleet of user embedding specialists achieves significant performance gains by leveraging diverse modeling strategies tailored to different facets of user behavior.
User representation is one of the highest-leverage modeling problems in industrial recommendation systems: a single advancement in how users are encoded can propagate across retrieval, ranking, and integrity tasks at platform scale. Prior industrial user representation work builds either a single user model that emits one or more embedding vectors or a shared backbone with task-specific adaptation. In this paper, we present Mosaic, a foundational user modeling platform that employs a fleet of specialists to learn user embeddings. The fleet comprises four architecturally diverse model families - memorization-driven, dense-heavy, sequential-based, and CoTrain models - each focusing on a distinct facet of user behavior. We developed MRM (Multi-task Relations Mining) and CRL (Cosine Redundancy Loss) techniques to maximize the marginal information contribution of each new specialist. We also introduce CoEval and User Tower Zero-Out, new logging-free embedding evaluation framework that improves development velocity while preserving downstream-aligned accuracy. Our hybrid CPU/GPU, online-and-offline serving stack allows each specialist to choose the adequate serving strategy to meet the freshness, latency, and computational requirements. Mosaic delivers consistent and significant offline NE improvements in addition to online gains.