GulkoA/TinyStories-tokenized-Llama-3.2
收藏Hugging Face2025-03-25 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/GulkoA/TinyStories-tokenized-Llama-3.2
下载链接
链接失效反馈官方服务：
资源简介：
TinyStories数据集是一个经过Llama-3.2模型tokenization的文本数据集，包含训练集和验证集，适用于加速稀疏自动编码器的训练和测试。数据集的上下文窗口大小为128，且数据未打乱。数据集总大小约为1.86亿字节，下载大小约为7936万字节。数据集可用于文本生成任务。
The TinyStories dataset is a text dataset tokenized with the Llama-3.2 model, including both training and validation sets, suitable for accelerating the training and testing of sparse autoencoders. The dataset has a context window size of 128 and the data is not shuffled. The total size of the dataset is approximately 1.86 billion bytes, with a download size of approximately 7.93 million bytes. The dataset can be used for text generation tasks.
提供机构：
GulkoA
