让大模型学会根据多模态信息修正时间序列预测结果。
Rethinking Post-Training Recipes for Multimodal Time-Series Forecasting
- 用监督微调与可验证奖励强化学习，训练大模型修正数值模型预测。
- 在TimesX数据集上，比纯数值模型和纯大模型方法提升显著。
- 适合需要融合文本、图像等非数值信息的预测场景。
时间序列基础模型（TSFMs）在零样本数值预测中表现优异，但无法处理影响真实轨迹的多模态非数值上下文。本文提出PostTime，一种后训练方法，通过监督微调（SFT）和可验证奖励强化学习（RLVR），使大语言模型（LLM）基于多模态上下文对强数值模型先验进行条件化修正。该方法生成自动化推理轨迹，指导模型决定是否修改、保留或忽略原始预测。我们在TimesX多模态预测基准上评估，使用Gemma-3-4B LLM与TimesFM-2.5 TSFM，结果表明其显著优于独立的TSFM、LLM基线及现有方法。
原文摘要 · Abstract (English)
Time-Series Foundation Models (TSFMs) excel at zero-shot unimodal forecasting using numerical data, but unlike LLMs they cannot consume multimodal, non-numerical context that often shape real-world trajectories. In this work, we bridge this gap and argue for a multimodal time-series forecasting approach that post-trains LLMs to act as context-guided revisors over strong numerical TSFM priors. We introduce PostTime, a post-training recipe combining Supervised Fine-Tuning (SFT) and Reinforcement Learning with Verifiable Rewards (RLVR), along with a methodology to generate automated reasoning traces for forecast revisions. PostTime teaches an LLM to generate context-conditioned forecast interventions -- decisions to revise, preserve, or ignore the TSFM prior based on the multimodal context. We evaluate this approach on the TimesX multimodal forecasting benchmark using a Gemma-3-4B LLM and TimesFM-2.5 TSFM, and show that it significantly outperforms standalone TSFMs, LLM-only baselines, and existing multimodal forecasting approaches.
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。