Vikhrmodels/AudioBooksInstructGemini2.5
收藏Hugging Face2025-12-15 更新2025-12-20 收录
下载链接:
https://hf-mirror.com/datasets/Vikhrmodels/AudioBooksInstructGemini2.5
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含来自俄语有声读物的录音,带有自动生成的问题和答案,用于训练模型遵循指令。数据集包含多种任务类型,如指令遵循、摘要、分类和命名实体识别(NER)。数据格式包括音频、文本、问题、答案、任务类型、对话和语音名称等字段。对话采用ChatML格式。数据集基于ToneBooks生成,使用Google Gemini 2.5 Flash Lite模型通过OpenRouter生成问题和答案。
The dataset contains audio recordings from Russian audiobooks with automatically generated questions and answers for training models to follow instructions. It includes various task types such as instruction following, summarization, classification, and named entity recognition (NER). The data format includes fields like audio, text, question, answer, task_type, conversations, and voice_name. Conversations are in ChatML format. The dataset is generated based on ToneBooks using the Google Gemini 2.5 Flash Lite model via OpenRouter to generate questions and answers.
提供机构:
Vikhrmodels



