telecomadm1145/skr_trans_distill

本数据集由SakuraLLM大模型生成机器翻译结果，主要用于模型蒸馏训练。它包含日文原文及其对应的中文翻译，适用于日译中任务的模型训练与蒸馏，特别是知识蒸馏场景下的小模型训练。数据集由telecomadm1145策划，语言为日语和中文，采用MIT许可证。原始日文文本来源于轻小说，翻译由SakuraLLM自动完成，无人工标注。数据集以Parquet格式存储，包含raw（日文原文）和translated（中文翻译）两列。需要注意的是，翻译质量受限于SakuraLLM模型能力，可能存在错误或偏差，不建议作为高质量人工翻译语料使用。

This dataset is generated by the SakuraLLM large model for machine translation results, primarily used for model distillation training. It contains Japanese original texts and their corresponding Chinese translations, suitable for model training and distillation in Japanese-to-Chinese translation tasks, especially for small model training in knowledge distillation scenarios. The dataset is curated by telecomadm1145, with languages including Japanese and Chinese, under the MIT license. The original Japanese texts are sourced from light novels, and translations are automatically completed by SakuraLLM without human annotation. The dataset is stored in Parquet format, including columns raw (Japanese original text) and translated (Chinese translation). It should be noted that translation quality is limited by the capabilities of the SakuraLLM model and may contain errors or biases; thus, it is not recommended for use as high-quality human translation data.