easyr1-v2-pro-apps-with-electron-data-plus-icon-data-from-yt-2k-4MP

# easyr1-v2-pro-apps-with-electron-data-plus-icon-data-from-yt-2k-4MP

This dataset was generated using the EasyR1 grounding dataset pipeline.

## Generation Details

- **Generated on**: 2025-09-03 08:32:38 UTC
- **Script**: `push_easyr1_to_hf.py`
- **Data directory**: `datasets`

## Parameters Used

- **Maximum samples**: 2000
- **Image resize (max megapixels)**: 4.0 MP
- **Minimum native image resolution**: 0.0 MP
- **Prompt format**: `gta1`
- **Output format**: `coordinates`
- **Random seed**: 42
- **Resampling enabled**: False
- **Icon upsampling ratio**: Disabled (random sampling)
- **pc-agent-e deduplication**: False
- **Debug images enabled**: True (annotated images with red bounding boxes included)

## Dataset Groups

The following JSON/JSONL files were used to create this dataset:

### Dataset Group 1
- datasets/grounding_annotations.json

### Dataset Group 2
- datasets/grounding_text_element_annotations_from_bbox_agreement.json

### Dataset Group 3
- datasets/powerpoint_shard_0000.normalized.json

## Dataset Statistics

- **Total training samples**: 2000
- **Image dimensions**: Variable
- **Columns**: image_path, prompt, normalized_bbox, images, easyr1_prompt, bbox, messages, original_image_width, image_width, image_height, annotated_images

## System Prompt

The following system prompt is used for this dataset:

```
You are an expert UI element locator. Given a GUI image and a user's element description, provide the coordinates of the specified element as a single (x,y) point. For elements with area, return the center point.

Output the coordinate pair exactly:
(x,y)
```

## Sample Entry

- **User prompt**: <image>select a different blending mode
- **Assistant response**: (1470,795)
- **Bounding box**: [1399, 790, 1541, 800]
- **Image path**: images/Photoshop 2025 Free Crash Course_scene0025_t001748.250.jpg

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("mlfoundations-cua-dev/easyr1-v2-pro-apps-with-electron-data-plus-icon-data-from-yt-2k-4MP")

# Access the training data
train_data = dataset['train']

# Example: Get the first sample
sample = train_data[0]
images = sample['images']
messages = sample['messages']
bbox = sample['bbox']
annotated_images = sample['annotated_images'] # Available when debug images are enabled```

## Prompt Formats

### gta1

Standard GTA1 format without resolution information. Outputs coordinates in (x,y) format.

## License

Please refer to the original dataset licenses for usage restrictions.

# easyr1-v2-pro-apps-with-electron-data-plus-icon-data-from-yt-2k-4MP

本数据集依托EasyR1基础数据集流水线构建生成。

## 生成详情
- **生成时间**：2025-09-03 08:32:38 UTC
- **脚本**：`push_easyr1_to_hf.py`
- **数据目录**：`datasets`

## 所用参数
- **最大样本数**：2000
- **图像缩放（最大像素数）**：4.0 MP
- **原生图像最小分辨率**：0.0 MP
- **提示词格式**：`gta1`
- **输出格式**：`coordinates`（坐标）
- **随机种子**：42
- **启用重采样**：False（未启用）
- **图标上采样比例**：已禁用（采用随机采样）
- **pc-agent-e去重**：False（未启用）
- **启用调试图像**：True（包含带有红色边界框的标注图像）

## 数据集分组
本次数据集构建使用了以下JSON/JSONL文件：

### 数据集分组1
- `datasets/grounding_annotations.json`

### 数据集分组2
- `datasets/grounding_text_element_annotations_from_bbox_agreement.json`

### 数据集分组3
- `datasets/powerpoint_shard_0000.normalized.json`

## 数据集统计信息
- **总训练样本数**：2000
- **图像尺寸**：可变
- **数据字段**：image_path、prompt、normalized_bbox、images、easyr1_prompt、bbox、messages、original_image_width、image_width、image_height、annotated_images

## 系统提示词
本数据集使用如下系统提示词：

你是一名专业的UI元素定位专家。给定图形用户界面（Graphical User Interface，GUI）图像与用户的元素描述，请输出指定元素的坐标点。对于存在面积的元素，返回其中心点坐标。请严格按照以下格式输出坐标对：
(x,y)

## 样本示例
- **用户提示词**：<image>选择其他混合模式
- **助手回复**：(1470,795)
- **边界框**：[1399, 790, 1541, 800]
- **图像路径**：images/Photoshop 2025 Free Crash Course_scene0025_t001748.250.jpg

## 使用方法
python
from datasets import load_dataset

# 加载目标数据集
dataset = load_dataset("mlfoundations-cua-dev/easyr1-v2-pro-apps-with-electron-data-plus-icon-data-from-yt-2k-4MP")

# 访问训练集数据
train_data = dataset['train']

# 示例：获取第一条样本
sample = train_data[0]
images = sample['images']
messages = sample['messages']
bbox = sample['bbox']
# 调试图像启用时可访问该字段
annotated_images = sample['annotated_images']

## 提示词格式
### gta1
标准GTA1格式，不含分辨率信息，输出采用(x,y)坐标格式。

## 许可证
使用限制请参考原始数据集的许可证条款。