Towards a Densing Law for User Representation Learning at Billion-Scale Capacity

Bin Dou, Junru Zhang, Zhaoyi Yuan, Wuliang Huang, Letian Gong, Baokun Wang, Huan Li, Yu Cheng, Weiqiang Wang

arXiv:2608.23392 · 2026-08-25 공개 · arXiv · PDF

tokenization representation-space large-scale scaling-law adaptive-tokenization large-scale-learning tokenized-scaling user-behavior

Abstract

User representation learning in real-world industrial scenarios is commonly scaled by increasing user amount, behavioral sequence length and model size. However, existing methods face two challenges: (i) Bottleneck for raw data scaling at billion-scale capacity, as performance exhibit diminishing performance gains with larger-scale raw text user behavioral input, which can be mitigated by tokenization. (ii) Lack of quantitative analysis of how tokenization configurations should scale with data size. In this report, we propose User Behavioral Densing Law for characterizing the quantitative relationship between data scale and the minimum sufficient tokenization capacity. Firstly, we conduct a pilot study on raw & tokenized scaling comparison on billion-scale Alipay dataset, revealing the raw data scaling bottleneck and the sustained gains enabled by tokenization. To derive the scaling pattern governing the minimum sufficient tokenization configuration at different data scales, theoretical analysis and systematic experiments are employed to summarize the quantitative scaling pattern. We find an approximately linear relationship between the logarithms of minimum sufficient tokenization capacity and input data size measured by tokens, and the scaling slope varies systematically with the tokenization method and data source, reflecting differences in representation-space redundancy and intra-source uniqueness. Guided by the proposed law, we further develop ALGN, an adaptive variable-length tokenization method that improves capacity allocation. Extensive experiments across diverse data sources, tokenization methods, and downstream tasks demonstrate the generalizability and reliability of the User Behavioral Densing Law, providing practical guidance for tokenization configuration selection in large-scale user representation learning. Moreover, ALGN outperforms existing baselines.

한국어 요약

한 줄 요약

유저 행동 데이터의 밀도 기반 토큰화 법칙(User Behavioral Densing Law)을 제안하고, ALGN이라는 적응형 토큰화 방법으로 대규모 유저 표현 학습 성능을 향상시킨다.

핵심 기여도

핵심 아이디어

유저 행동 데이터는 반복적이고 정보 밀도가 낮아, 단순히 데이터 양이나 모델 크기를 늘리는 방식은 성능 향상에 한계가 있다. 이에 따라, **Behavioral Densing Law**는 데이터 규모와 최소한의 토큰화 용량 간의 관계를 정량적으로 정의하며, 토큰화를 통해 정보 밀도를 높이는 것이 핵심이다.

**ALGN**은 **RQ-VAE**(Residual Quantization VAE) 기반의 **변수 길이 토큰화**를 통해, 유저 행동 복잡도에 따라 토큰 길이를 조절한다. 이는 **잔차 노름**(residual norm)과 **코드 불확실도**(code uncertainty)를 기반으로 각 샘플에 대해 필요한 토큰 수를 결정하는 **로컬 마진 유틸리티**(local marginal utility) 원칙을 구현한 것이다.

기술적 접근법

주요 결과

의의 및 한계

실용적 활용