TinyCast：具計算週期性之機率性零樣本預測

摘要

我們提出 TinyCast，一個無注意力機制的零樣本預測器，僅以 146,505 個參數即可輸出預測分佈。其設計前提在於：在此規模下，上下文的週期結構值得計算而非學習。一個零參數頻譜檢測器提供主導週期，上下文按其相位進行摺疊，再由膨脹卷積編碼器與區塊自迴歸分位數解碼器處理其餘結構。在 GIFT-Eval 排行榜中，凡是可確認參數量的零樣本參賽模型，TinyCast 均比其更小。就機率性準確度而言，它定義了規模-精度前沿。在聲明無測試資料洩漏的零樣本參賽模型中，它是唯一低於 1.4M 參數且能輸出預測分佈的模型；所有得分更高的參賽模型至少都具備該參數量。在 Chronos-ZS 與 fev-bench 上，所有領先於它的神經模型，其參數量至少為它的 28 倍。由於其混合路徑僅包含卷積與矩陣乘法，因此可匯出為靜態 INT8 格式，並在嵌入式裝置上進行端到端預測，無需逐訊號擬合。

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.