TinyCast：基于计算周期性的概率零样本预测

摘要

我们提出TinyCast，一种无注意力机制的零样本预测器，仅用146,505个参数即可输出预测分布，其设计前提是：在此规模下，上下文的周期结构值得显式计算而非学习获取。一个零参数频谱检测器提供主导周期，上下文按其相位折叠，随后由膨胀卷积编码器和块自回归分位数解码器对剩余部分建模。在GIFT-Eval榜单上，它是所有可确定参数量的零样本参赛模型中规模最小的。在概率精度上，它定义了规模-精度前沿。在声明无测试数据泄漏的零样本参赛模型中，它是唯一参数低于140万的能输出预测分布者，而所有得分更高的参赛模型至少拥有该参数量。在Chronos-ZS和fev-bench上，每一个排名在其之前的神经模型参数量至少为其28倍。由于其混合路径仅包含卷积和矩阵乘法，该模型可导出为静态INT8格式，并在嵌入式设备上端到端执行预测，无需逐信号拟合。

English

We introduce TinyCast, an attention-free zero-shot forecaster that emits a predictive distribution from 146,505 parameters, on the premise that at this size the periodic structure of a context is worth computing rather than learning. A zero-parameter spectral detector supplies the dominant periods, the context is folded on their phase, and a dilated convolutional encoder and a block-autoregressive quantile decoder model the rest. It is smaller than every zero-shot entry on the GIFT-Eval board whose parameter count can be established. On probabilistic accuracy it defines the size-accuracy frontier. Among zero-shot entries declaring no test-data leakage it is the only one below 1.4M parameters that emits a predictive distribution, and every entry scoring better carries at least that budget. On Chronos-ZS and fev-bench every neural model ahead of it carries at least 28 times its parameters. Because the mixing path is convolutions and matrix multiplications only, it exports to static INT8 and forecasts end to end on an embedded device without per-signal fitting.