Search papers, labs, and topics across Lattice.
Timer-S1, a Mixture-of-Experts time series foundation model with 8.3B parameters, achieves state-of-the-art forecasting performance by employing Serial Scaling across model architecture, dataset, and training pipeline. The model integrates sparse TimeMoE and TimeSTP blocks for Serial-Token Prediction (STP), a training objective designed to improve long-term predictions. A post-training stage, including continued pre-training and long-context extension, further enhances Timer-S1's performance on both short-term and long-context forecasting tasks.
Time series forecasting gets a serious upgrade: Timer-S1, a new foundation model, blows away benchmarks by scaling model size, dataset size, and training in a coordinated "serial" fashion.
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 pre-training and long-context extension, to enhance short-term and long-context performance. Evaluated on the large-scale GIFT-Eval leaderboard, Timer-S1 achieves state-of-the-art forecasting performance, attaining the best MASE and CRPS scores as a pre-trained model. Timer-S1 will be released to facilitate further research.
