[2603.04791] Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling

Computer Science > Artificial Intelligence arXiv:2603.04791 (cs) [Submitted on 5 Mar 2026] Title:Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling Authors:Yong Liu, Xingjian Su, Shiyu Wang, Haoran Zhang, Haixuan Liu, Yuxuan Wang, Zhou Ye, Yang Xiang, Jianmin Wang, Mingsheng Long View a PDF of the paper titled Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling, by Yong Liu and 9 other authors View PDF HTML (experimental) Abstract:We introduce Timer-S1, a strong Mixture-of-Experts (MoE) time series foundation model with 8.3B total parameters, 0.75B activated parameters for each token, and a context length of 11.5K. To overcome the scalability bottleneck in existing pre-trained time series foundation models, we perform Serial Scaling in three dimensions: model architecture, dataset, and training pipeline. Timer-S1 integrates sparse TimeMoE blocks and generic TimeSTP blocks for Serial-Token Prediction (STP), a generic training objective that adheres to the serial nature of forecasting. The proposed paradigm introduces serial computations to improve long-term predictions while avoiding costly rolling-style inference and pronounced error accumulation in the standard next-token prediction. Pursuing a high-quality and unbiased training dataset, we curate TimeBench, a corpus with one trillion time points, and apply meticulous data augmentation to mitigate predictive bias. We further pioneer a post-training stage, including continued ...