大模型小增益：冻结预测模型在MONSTER上的分类评测
原标题：Big Models, Small Gains: Frozen Forecasters as Classifiers on MONSTER
AI 摘要
研究对两个时间序列预测基础模型 Reverso 和 Chronos-Bolt 在 MONSTER 基准的六个大型数据集上进行受控评测，使用 1%、10%、100% 三种标签预算，将其冻结表示与四种强分类基线及同架构随机权重版本对比。结果显示预训练相比随机权重在六个数据集中的五个上提升分类准确率，最多提升 23 个百分点，但冻结表示在 17 种设置中有 15 种不如最强非预训练基线、2 种持平。恢复振幅信息可部分缩小差距，微调和预测误差特征无法弥合差距，作者据此强调评估预测基础模型时需使用强任务特定基线和随机权重对照。
正文节选
Big Models, Small Gains: Frozen Forecasters as Classifiers on MONSTER Abstract Forecasting foundation models are increasingly used as frozen feature extractors for time series classification, but it remains unclear whether their pretrained representations offer an advantage over strong classifiers that do not require pretraining. This raises an important question: do these representations provide useful information beyond what the model architecture alone can capture, particularly when labelled