ayarlicazhocam_finetune
收藏资源简介:
Ayarlicaz Hocam Fine-Tuning Dataset 是一个为大型语言模型微调设计的双语指令遵循数据集,包含英语和土耳其语两个版本。该数据集专门用于监督微调,包含对话式指令-响应对,适用于指令遵循模型训练、提示工程实验、LLM研究、聊天机器人开发和教育演示等多种应用场景。数据集采用标准的指令数据集格式,每个样本包含instruction、input和output三个字段,其中input字段通常为空字符串。数据集总规模为2,009个样本(英语1,005个,土耳其语1,004个),数据大小为2.4MB。数据集结构为消息列表格式,每个消息包含内容、图像、角色、思考和工具调用等字段。该数据集由Görkem Ergüne创建,采用MIT许可证发布。
Ayarlicaz Hocam Fine-Tuning Dataset is a bilingual instruction-following dataset designed for large language model (LLM) fine-tuning, with both English and Turkish versions. This dataset is specifically developed for supervised fine-tuning, containing conversational instruction-response pairs, and is suitable for multiple application scenarios including instruction-following model training, prompt engineering experiments, LLM research, chatbot development, and educational demonstrations. The dataset follows the standard instruction dataset format, where each sample includes three fields: instruction, input, and output, with the input field usually being an empty string. The total scale of the dataset is 2,009 samples (1,005 in English and 1,004 in Turkish), with a data size of 2.4 MB. The dataset is structured as a message list format, where each message contains fields such as content, image, role, thought, and tool call. This dataset was created by Görkem Ergüne and released under the MIT License.
数据集概述
- 数据集名称: Ayarlıcaz Hocam Fine-Tuning Dataset
- 数据集地址:
https://huggingface.co/datasets/gorkemergune/ayarlicazhocam_finetune - 语言: 英语(en)、土耳其语(tr)
- 许可证: MIT
- 任务类别: 文本生成(text-generation)
数据集详情
- 数据集大小: 约 4.83 MB(下载大小约 3.97 MB)
- 样本总数: 2009 条
- 英语(en): 1005 条
- 土耳其语(tr): 1004 条
- 训练集(train): 2009 条(包含所有语言)
- 数据分割: 提供
train、tr、en三个子集
数据格式
- 特征字段:
messages列表,包含:content: 字符串类型(对话内容)role: 字符串类型(角色,如用户或助手)thinking: 字符串类型(思考过程)images: null(无图像)tool_calls: null(无工具调用)
- 示例结构:
json { "instruction": "User instruction", "input": "", "output": "Assistant response" }
用途
- 监督微调(SFT)
- 指令跟随模型训练
- 大语言模型(LLM)研究
- 提示工程实验
- 聊天机器人开发
- 基准测试
- 教育演示
使用方式
-
加载数据集(Python): python from datasets import load_dataset dataset = load_dataset("gorkemergune/ayarlicazhocam_finetune")
-
配置文件: 默认配置(default),支持
train、tr、en三个数据文件分割
作者
- 名称: Görkem Ergüne
- GitHub:
https://github.com/gorkemergune - Hugging Face:
https://huggingface.co/gorkemergune




