用真实中断数据训练大模型，提升供应链风险预测准确率。
Forecasting Supply Chain Disruptions with Foresight Learning
- 用实际中断结果监督大模型，生成校准的概率预测
- 性能超越GPT-5等基线，精度与校准度显著提升
- 无需额外提示即可生成结构化推理，适合决策支持
提前预判供应链中断是企业与政策制定者的核心挑战。主要难点在于：从嘈杂、非结构化的输入中学习对罕见但高影响事件的可靠推理，通用模型在此类任务中缺乏适应性。本文提出一个端到端框架，利用实际发生的中断结果作为监督信号，训练大语言模型生成校准的概率预测。结果表明，该模型在准确性、校准度和精确度上均显著优于强基线（包括GPT-5）。此外，训练过程促使模型自发产生更结构化、可靠的概率推理能力，无需显式提示。这些发现揭示了一条构建领域专用预测模型的通用路径，可输出可供决策使用的信号。为保障透明性，本文开源了评估数据集。数据集：https://huggingface.co/datasets/LightningRodLabs/supply-chain-predictions
原文摘要 · Abstract (English)
Anticipating supply chain disruptions before they materialize is a core challenge for firms and policymakers alike. A key difficulty is learning to reason reliably about infrequent, high-impact events from noisy and unstructured inputs - a setting where general-purpose models struggle without task-specific adaptation. We introduce an end-to-end framework that trains LLMs to produce calibrated probabilistic forecasts using realized disruption outcomes as supervision. The resulting model substantially outperforms strong baselines - including GPT-5 - on accuracy, calibration, and precision. We also show that training induces more structured and reliable probabilistic reasoning without explicit prompting. These results suggest a general pathway for training domain-specific forecasting models that produce decision-ready signals. To support transparency we open-source the evaluation dataset used in this study. Dataset: https://huggingface.co/datasets/LightningRodLabs/supply-chain-predictions
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。