captcha数据使用说明
数据概览
本数据集包含8类专用验证码数据,每类任务包含2k条数据,用于训练多模态验证码识别模型。
| 任务类型 | SFT数据文件 | 图像数据目录 |
|---|---|---|
| 图片连接 | captcha_connect_icon.jsonl |
captcha/connect_icon |
| 按序点击 | click_order_dataset.jsonl |
captcha/click_order |
| 坐标定位 | coordinates_dataset.jsonl |
captcha/coordinates |
| 飞镖计数 | dart_counting_dataset.jsonl |
captcha/dart_counting |
| 骰子计数 | dice_counting_dataset.jsonl |
captcha/dice_counting |
| 图像选择 | image_select_dataset.jsonl |
captcha/image_select |
| 区域选择 | patch_select_dataset.jsonl |
captcha/patch_select |
| 旋转匹配 | rotation_match_dataset.jsonl |
captcha/rotation_match |
数据结构说明
文件组织形式
- 图像文件夹:每个样本使用独立文件夹管理
- 样本文件夹内容:
- SFT训练所需的图片文件
- 一个JSON配置文件,包含:
- 各按钮的位置信息
- 训练标签值
- 测试时的辅助信息
注意:如果仅用于模型训练,可以忽略JSON配置文件。该文件主要用于SFT数据制作和测试集信息提供。
TuningFactory配置示例
在TuningFactory中使用该数据集进行SFT训练时,dataset_info.json 的配置方式如下:
"captcha_image_select": {
"file_name": "/data/cpfs_0/wangpengcheng/data/image_select_dataset.jsonl",
"formatting": "sharegpt",
"columns": {
"messages": "conversations",
"system": "system",
"image": "image"
},
"tags": {
"role_tag": "role",
"content_tag": "content",
"user_tag": "user",
"assistant_tag": "assistant"
}
}
使用建议
- 数据路径:请根据实际存储位置修改配置文件中的
file_name路径 - 批处理:可以根据计算资源情况调整批处理大小和训练策略