WDLMPro

WDLMPro是由信息与语言处理中心创建的一个用于评估语言模型对词汇理解能力的数据集。该数据集包含数千个synset组，每个组由一个目标词及其定义以及其分类姐妹词及其定义组成。数据集的构建基于WordNet，通过特定的语义关系来组织synset。WDLMPro主要用于测试语言模型在匹配词与其定义方面的能力，特别是在区分不同定义和词汇的细微差别上。该数据集的应用领域包括自然语言处理和语言模型研究，旨在解决现有模型在词汇理解上的局限性。

WDLMPro is a dataset developed by the Center for Information and Language Processing for evaluating the lexical comprehension abilities of language models. It consists of thousands of synset groups, with each group containing a target word and its definition, alongside its categorical sister terms and their respective definitions. The dataset is built upon WordNet, where synsets are organized using specific semantic relations. WDLMPro is primarily designed to test language models' capacity to match words with their definitions, especially in discerning subtle discrepancies between different definitions and lexical items. Its application domains cover natural language processing and language model research, with the goal of addressing the limitations of existing models in lexical understanding.