ckverma/snli_random
收藏Hugging Face2024-12-02 更新2024-12-14 收录
下载链接：
https://hf-mirror.com/datasets/ckverma/snli_random
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是SNLI数据集的一个变体，包含了随机触发器。数据集分为7个部分，包括训练集和验证集的多个子集。训练集从SNLI数据集的训练集中随机抽取6000个样本，其中3000个样本保持不变，另外3000个样本的前提假设前添加了随机触发器。验证集部分则分别从SNLI数据集的验证集中随机抽取1000个蕴含、中立和矛盾的样本，并为每个类别生成一个添加了随机触发器的子集。
This dataset consists of 7 splits: train, entailment, entailment_triggers, neutral, neutral_triggers, contradiction, and contradiction_triggers. The train split contains 6,000 samples, with 3,000 unchanged and the other 3,000 having random triggers prepended to the hypothesis. The other splits, such as entailment, neutral, and contradiction, are randomly selected from the validation split of the SNLI dataset and have random triggers prepended to them.
提供机构：
ckverma
