重新思考稠密检索的少样本能力（英文版）

英文标题：Rethinking Dense Retrieval's Few-Shot Ability中文摘要：本文提出 FewDR 数据集和统一评估基准，并消除了现有方法中出现的一些问题。经实证结果表明，当前最先进的 DR 模型仍然在标准 few-shot 场景中面临挑战。英文摘要：Few-shot dense retrieval (DR) aims to effectively generalize to novel searchscenarios by learning a few samples. Despite its importance, there is littlestudy on specialized datasets and standardized evaluation protocols. As aresult, current methods often resort to random sampling from superviseddatasets to create "few-data" setups and employ inconsistent t