shadowpa0327/qwen3_8b_eagle3-parquet

qwen3_8b_eagle3数据集是一个基于Parquet格式的对话数据集，源自Tengyunw/qwen3_8b_eagle3的原始JSON文件。该数据集被转换为61个Parquet分片，每个分片约10,000行，以支持通过datasets库进行流式友好访问。数据遵循ShareGPT格式，包含id和conversations字段，其中conversations是一个列表结构，包含from和value字段，用于表示对话中的发言者和内容。数据集总共有607,865行对话数据，使用zstd压缩，适用于自然语言处理任务，如对话生成、模型训练或分析。

The qwen3_8b_eagle3 dataset is a conversational dataset in Parquet format, derived from the original JSON file of Tengyunw/qwen3_8b_eagle3. It is converted into 61 Parquet shards, each containing approximately 10,000 rows, to enable streaming-friendly access via the datasets library. The data adheres to the ShareGPT format, with fields including id and conversations, where conversations is a list of structs with from and value fields to represent the speaker and content in dialogues. The dataset comprises a total of 607,865 rows of conversational data, compressed using zstd, and is suitable for natural language processing tasks such as dialogue generation, model training, or analysis.