jihuny/pythia1.4b_10k_triplet_unnorm_penul_prev
收藏Hugging Face2025-11-03 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/jihuny/pythia1.4b_10k_triplet_unnorm_penul_prev
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含多个特征字段，如提示文本(prompt)、索引(prompt_idx)、选择的文本(chosen)、被拒绝的文本(rejected)及其索引，以及各种分数和完成原因等。数据集分为训练集(train)，共有450000个样本，整个数据集的大小为8157592071字节。此外，还提供了默认配置信息和数据文件的路径。
The dataset includes multiple feature fields such as prompt text (prompt), index (prompt_idx), chosen text (chosen), rejected text (rejected) and their indices, as well as various scores and finish reasons. The dataset is split into a training set (train) with a total of 450,000 samples, and the entire dataset size is 8157592071 bytes. In addition, default configuration information and data file paths are provided.
提供机构：
jihuny
