izzako/javanese-pixelgpt
收藏Hugging Face2025-12-18 更新2025-12-20 收录
下载链接：
https://hf-mirror.com/datasets/izzako/javanese-pixelgpt
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含用于训练PixelGPT模型的预处理爪哇语文本数据。数据集语言为爪哇语，总样本数为401,542个，其中训练样本400,726个，测试样本816个。数据集特征包括文档标识符、块标识符、像素值、基于字素的标记ID、LLaMA标记ID和原始文本块。还提供了渲染器配置和使用示例。
This dataset contains preprocessed Javanese text data for training PixelGPT models. The dataset language is Javanese, with a total of 401,542 samples, including 400,726 training samples and 816 test samples. Dataset features include document identifiers, chunk identifiers, pixel values, grapheme-based token IDs, LLaMA token IDs, and original text chunks. Renderer configuration and usage examples are also provided.
提供机构：
izzako
