更小更快的时序预测模型，用简化架构提升精度与效率
Moirai 2.0: When Less Is More for Time Series Forecasting
- 采用单输入、量化损失的解码器架构，摒弃复杂训练方式
- 比前代快2倍、小30倍，且在多个数据集上表现更优
- 适合追求高效部署与高精度预测的工业级应用
我们提出Moirai 2.0，一个基于3600万条时序数据训练的解码器仅架构时序基础模型。该模型采用分位数预测与多标记输出机制，在概率准确性与推理效率之间取得更好平衡。在Gift-Eval基准上，其性能位居预训练模型前列，同时实现准确率、速度与模型规模的优良权衡。相比Moirai 1.0，Moirai 2.0舍弃掩码编码器训练、多片段输入与混合分布输出，改用更简单的解码器架构、单片段输入和分位数损失。消融实验表明，解码器主干与递归分位数解码是性能提升的核心。额外实验显示，它优于同家族更大模型，并具备稳健的跨领域表现。在效率与规模方面，Moirai 2.0比其最佳版本Moirai 1.0-Large快2倍、小30倍，且性能更优。模型性能随参数量增加趋于饱和，长序列预测性能下降，提示未来需加强数据扩展与长程建模。代码与评估细节已公开，支持后续研究。
原文摘要 · Abstract (English)
We introduce Moirai 2.0, a decoder-only time-series foundation model trained on a new corpus of 36M series. The model adopts quantile forecasting and multi-token prediction, improving both probabilistic accuracy and inference efficiency. On the Gift-Eval benchmark, it ranks among the top pretrained models while achieving a strong trade-off between accuracy, speed, and model size. Compared to Moirai 1.0, Moirai 2.0 replaces masked-encoder training, multi-patch inputs, and mixture-distribution outputs with a simpler decoder-only architecture, single patch, and quantile loss. Ablation studies isolate these changes -- showing that the decoder-only backbone along with recursive multi-quantile decoding contribute most to the gains. Additional experiments show that Moirai 2.0 outperforms larger models from the same family and exhibits robust domain-level results. In terms of efficiency and model size, Moirai 2.0 is twice as fast and thirty times smaller than its prior best version, Moirai 1.0-Large, while also performing better. Model performance plateaus with increasing parameter count and declines at longer horizons, motivating future work on data scaling and long-horizon modeling. We release code and evaluation details to support further research.
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。