llava-med-zh-instruct-60k

This Chinese dataset was translated from [llava-med](https://github.com/microsoft/LLaVA-Med) using [Qwen1.5-14B-Chat](https://huggingface.co/Qwen/Qwen1.5-14B-Chat) and contains 60k medical visual instruction data points.

You can organize content in the dataset_info.json in [LLaMA Factory](https://github.com/hiyouga/LLaMA-Factory) like this:
```
"llava_med_zh_60k": {
"hf_hub_url": "BUAADreamer/llava-med-zh-instruct-60k",
"formatting": "sharegpt",
"columns": {
"messages": "messages",
"images": "images"
},
"tags": {
"role_tag": "role",
"content_tag": "content",
"user_tag": "user",
"assistant_tag": "assistant"
}
}
```

Then you can use it in [LLaMA Factory](https://github.com/hiyouga/LLaMA-Factory) by specifying --dataset llava_med_zh_60k.

本中文数据集源自[LLaVA-Med](https://github.com/microsoft/LLaVA-Med)，依托[Qwen1.5-14B-Chat](https://huggingface.co/Qwen/Qwen1.5-14B-Chat)完成翻译，共包含6万条医学视觉指令数据样本。

你可以按照如下格式在[LLaMA Factory](https://github.com/hiyouga/LLaMA-Factory)的dataset_info.json中配置数据集内容：

"llava_med_zh_60k": {
"hf_hub_url": "BUAADreamer/llava-med-zh-instruct-60k",
"formatting": "sharegpt",
"columns": {
"messages": "messages",
"images": "images"
},
"tags": {
"role_tag": "role",
"content_tag": "content",
"user_tag": "user",
"assistant_tag": "assistant"
}
}

随后你在使用[LLaMA Factory](https://github.com/hiyouga/LLaMA-Factory)时，仅需通过指定`--dataset llava_med_zh_60k`参数即可调用该数据集。

搜集汇总

背景与挑战

背景概述

llava-med-zh-instruct-60k是一个包含60k条中文医疗视觉指令数据的数据集，适用于LLaMA Factory，由llava-med数据集翻译而来。