dlion168/ppl_test
收藏官方服务：
资源简介：
该数据集包含与文本生成任务相关的多个字段，如正负样本的token-wise损失，模型生成的延续tokens和音频，以及一些辅助性的数值特征。数据集分为训练集，包含1600个示例，数据大小为709MB。
The dataset includes multiple fields related to text generation tasks, such as token-wise loss for positive and negative samples, model-generated continuation tokens and audio, as well as some auxiliary numerical features. The dataset is split into a training set containing 1600 examples, with a total size of 709MB.
提供机构：
dlion168
