发表机构
The University of Melbourne; Monash University（墨尔本大学; 莫纳什大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究提出 SCALER 框架，通过轻量 Transformer 预测未来粗粒度形状引导 LLM 迭代精调，在多类预测任务中优于基线且降低推理成本。
AI 中文摘要
长期时间序列预测得益于趋势、季节性等全局结构的保留。近期基于大语言模型（LLM）的预测器常通过测试时缩放（如迭代精调）提升精度，但这些方法计算成本高，且随预测 horizon 延长，更易出现全局形状不匹配问题。我们提出 SCALER，一种由粗到精的预测框架：首先采用专为长期形状建模定制的轻量 Transformer，预测未来动态的粗粒度表示；该预测形状随后作为紧凑引导，供 LLM 通过迭代粗到精残差 token 精调执行测试时缩放，且每一步处理的 token 数量大幅减少。通过用显式未来形状预测引导精调，SCALER 降低了对长描述提示的依赖，其固定步长精调避免了基于代价高昂的奖励模型的选择，进一步降低了计算开销。实验结果表明，SCALER 在长期、短期及零样本预测中均优于强预测基线，同时显著降低了与缩放 LLM 用于时间序列预测相关的推理成本。代码：this https URL。
英文摘要
Long-term time series forecasting benefits from preserving global structure such as trends and seasonality. Recent LLM-based forecasters often improve accuracy through test-time scaling (e.g., iterative refinement), but these methods are computationally expensive and increasingly prone to global-shape mismatch as the prediction horizon extends. We propose SCALER, a coarse-to-fine forecasting framework that first employs a lightweight Transformer tailored to long-term shape modeling to predict a coarse representation of future dynamics. This predicted shape then serves as a compact guide for an LLM to perform test-time scaling via iterative coarse-to-fine residual token refinement, while processing substantially fewer tokens at each step. By guiding refinement with an explicit future-shape prediction, SCALER reduces reliance on long description prompts, and its fixed-step refinement avoids costly reward-model-based selection, further lowering computational overhead. Experimental results demonstrate that SCALER outperforms strong forecasting baselines in long-term, short-term and zero-shot forecasting while significantly reducing the inference cost associated with scaled LLM for time series forecasting. Code: https://github.com/xuanmay2701/SCALER.
CommentsAccepted at KDD 2026 (Oral)
