斯坦福自然语言推理（SNLI）语料库

SNLI语料库（1.0版）是一个570k人工编写的英语句子对的集合，这些句子对经过手动标记以实现平衡分类，并带有enume，矛盾和中性标签，支持自然语言推理（NLI）的任务，也称为识别文本蕴涵（RTE）。可用于探索知识图谱推理、长短期记忆神经网络。数据集来自斯坦福自然语言处理小组 。

The SNLI (Stanford Natural Language Inference) Corpus (Version 1.0) is a dataset consisting of 570,000 manually written English sentence pairs. These pairs were manually annotated to achieve balanced classification, with three labels: entailment, contradiction, and neutral, to support the natural language inference (NLI) task, which is also known as recognizing textual entailment (RTE). This corpus can be used to explore knowledge graph reasoning and long short-term memory neural networks. The dataset is sourced from the Stanford Natural Language Processing Group.

搜集汇总

背景与挑战

背景概述

斯坦福自然语言推理（SNLI）语料库1.0版包含57万个人工标注的英文句子对，每个句子对带有蕴涵、矛盾或中性标签。该数据集用于自然语言推理（即文本蕴涵识别）任务，可作为评估文本表示系统的基准，数据来源于斯坦福自然语言处理小组。