habanoz/haber-90k-gpt-text

`habanoz/news-tr-90k`数据集包含土耳其语文本，分为训练集和验证集。数据集的列被合并成一个单一的文本文档，格式包括标题、摘要和内容。在该数据集上训练了一个具有8K词汇量的分词器，训练集包含62748074个token，验证集包含7015414个token。

The habanoz/news-tr-90k dataset contains Turkish news text, divided into training and validation sets. The dataset features include title, summary, and content. The training set has 81708 samples, and the validation set has 9079 samples. The dataset is used to train a tokenizer with an 8K vocabulary, containing 62748074 training tokens and 7015414 validation tokens.