Toto 2.0证明时间序列模型可规模化，5个版本覆盖400万到25亿参数。
Toto 2.0: Time Series Forecasting Enters the Scaling Era
- 统一训练配方下，模型从400万到25亿参数持续提升预测性能。
- 在BOOM、GIFT-Eval和TIME三个基准上均达新最优，尤其抗污染能力突出。
- 开源5个基础模型，适合需要高精度时序预测的研究与工程应用。
我们证明了时间序列基础模型具备可扩展性：仅用单一训练配方，模型性能随参数量从400万增至25亿而稳定提升。本文发布Toto 2.0系列，包含五个基于该配方训练的开源权重预测模型。该系列在三个预测基准上取得新最佳表现：自建的可观测性基准BOOM、标准通用基准GIFT-Eval，以及近期推出的抗污染基准TIME。报告详述实验结果及Toto 2.0的设计细节，包括架构、训练配方、训练数据和u-muP超参数迁移流程。所有五个基础检查点均已开放，采用Apache 2.0许可。
原文摘要 · Abstract (English)
We show that time series foundation models scale: a single training recipe produces reliable forecast-quality improvements from 4M to 2.5B parameters. We release Toto 2.0, a family of five open-weights forecasting models trained under this recipe. The Toto 2.0 family sets a new state of the art on three forecasting benchmarks: BOOM, our observability benchmark; GIFT-Eval, the standard general-purpose benchmark; and the recent contamination-resistant TIME benchmark. This report describes our experimental results and details the design decisions behind Toto 2.0: its architecture and training recipe, training data, and the u-muP hyperparameter transfer pipeline. All five base checkpoints are released under Apache 2.0.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。