AI 中文总结
该研究针对时间序列基础模型训练分布控制不足的问题，提出ORBIT训练范式，结合多级采样与增量训练，训练Falcon-2.0模型并引入Rank引导跨深度对齐，在多基准测试中展现优异零样本预测性能。
AI 中文摘要
时间序列基础模型（TSFMs）的进步主要依赖架构创新，而针对大规模异构语料库的训练机制仍未得到充分探索。因此，预训练分布在领域不平衡、上下文需求、预测 horizon 及缺失值方面往往控制不佳。我们提出ORBIT（全范围自举增量训练，Omni-Range Bootstrap Incremental Training），这一训练范式可使该分布明确且可控。ORBIT结合了自举多级采样（Bootstrap Multi-Level Sampling，用于控制数据集曝光并对记录、目标变量、上下文窗口和预测 horizon 进行采样）与全范围增量训练（Omni-Range Incremental Training，在单个训练阶段内改变上下文长度和预测 horizon）。在ORBIT框架下，我们训练了Falcon-2.0，这是一款简单的单变量仅编码器Transformer，具备感知缺失值的三通道patch标记化和并行patch预测能力。我们还提出了Rank引导跨深度对齐（Rank-Guided Cross-Depth Alignment），这一训练目标利用深层表示作为浅层的停止梯度教师，且无额外推理成本。在GIFT-Eval和fev-bench上的评估表明，其在不同领域和频率下具备出色的零样本预测性能。
英文摘要
Time series foundation models (TSFMs) have advanced primarily through architectural innovation, while training regimes for large-scale heterogeneous corpora remain under-explored. As a result, pre-training distributions are often poorly controlled with respect to domain imbalance, context requirements, prediction horizons, and missingness. We introduce ORBIT (Omni-Range Bootstrap Incremental Training), a training paradigm that makes this distribution explicit and controllable. ORBIT combines Bootstrap Multi-Level Sampling, which controls dataset exposure and samples records, target variables, context windows, and prediction horizons, with Omni-Range Incremental Training, which varies context lengths and prediction horizons throughout a single training stage. Under ORBIT, we train Falcon-2.0, a simple univariate encoder-only Transformer with missingness-aware triple-channel patch tokenization and parallel patch prediction. We further introduce Rank-Guided Cross-Depth Alignment, a training objective that uses late-layer representations as stop-gradient teachers for shallow layers without additional inference cost. Evaluations on GIFT-Eval and fev-bench demonstrate strong zero-shot forecasting performance across diverse domains and frequencies.
