SlimPer: Make Personalization Model Slim and Smart

2026-07-13

AI Summary

This work addresses the inefficiency of conventional Transformer-based recommender systems, which inherit generative architectures that incur linear growth in computation and memory with user history length and fail to align with the single-scoring nature of recommendation tasks. The authors propose a novel architecture that reframes personalized ranking as iterative optimization over a compact, unified user-item knowledge base. Each layer selectively queries multimodal user tokens with O(N) complexity and explicitly computes relevance scores, thereby decoupling model depth from history length. The approach employs a fixed-size intermediate representation and a request-level optimization mechanism to uniformly handle sparse, dense, and sequential features, while leveraging attention for intrinsic interpretability. Deployed in Instagram Reels and Feed, the system significantly boosts user engagement, efficiently supports modeling of histories exceeding ten thousand fine-grained interactions, and simplifies the overall architecture.

Abstract

Transformer-style architectures are increasingly adopted for industrial recommendation systems, yet they inherit a design premise misaligned with the task: generative models rely on per-token autoregressive prediction, which justifies maintaining large intermediate tensors that scale with sequence length. In contrast, recommendation systems produce a single set of relevance scores for each <user, item> pair without token-level supervision. Leveraging this observation, we propose SlimPer, which reformulates personalized ranking as iterative refinement of a compact, unified <user, item> knowledge base. At each layer, the model selectively queries raw multi-modal user-side tokens, computes explicit relevance matching scores, and refines the knowledge base, all in O(N) per-layer cost with a fixed-size intermediate representation. As a result, model depth is decoupled from user history length, enabling deeper relevance understanding without proportional growth in compute or memory; request-only optimization further trims memory by sharing a single copy of user-side tokens across all candidate items. SlimPer unifies sparse, dense, and sequence features within a single backbone and provides inherent interpretability through its attention mechanism. Deployed on Instagram Reels and Feed, SlimPer yields measurable improvements in user engagement while streamlining the overall system and enabling effective modeling of 10k+ fine-grained user history events.

Problem

• recommendation systems
• Transformer architecture
• personalization
• computational efficiency
• user history modeling

Innovation

• SlimPer
• recommendation systems
• iterative refinement
• fixed-size representation
• memory-efficient modeling