LivNLP/C-STS-Reannotated

这是一个针对条件语义文本相似度（C-STS）任务的重新注释数据集，包含了重新注释的训练集和验证集，用于评估模型在特定上下文条件下判断语义相似性的性能。数据集包含两个字段：sentence1和sentence2，表示句子对；condition字段表示特定的语义关注或上下文条件；label字段是相似度评分，范围从1到5，5表示最大语义等价。验证集中的-1标签表示无效实例。

This is a re-annotated dataset for Conditional Semantic Textual Similarity (C-STS) task, including re-annotated training and validation sets designed to evaluate the models performance in assessing semantic similarity under specific contextual conditions. The dataset contains two fields: sentence1 and sentence2 representing the sentence pair; the condition field represents the specific semantic focus or contextual condition; the label field is the similarity rating ranging from 1 to 5, with 5 indicating maximum semantic equivalence. A label of -1 in the validation set indicates an invalid instance.