| --- |
| language: |
| - km |
| license: cc-by-4.0 |
| task_categories: |
| - image-to-text |
| - visual-question-answering |
| tags: |
| - khmer |
| - ocr |
| - vlm |
| - surya-2 |
| - document-understanding |
| dataset_info: |
| features: |
| - name: id |
| dtype: string |
| - name: image |
| dtype: image |
| - name: messages |
| list: |
| - name: role |
| dtype: string |
| - name: content |
| dtype: string |
| splits: |
| - name: train |
| num_bytes: 31424082 |
| num_examples: 99 |
| download_size: 31428464 |
| dataset_size: 31424082 |
| configs: |
| - config_name: default |
| data_files: |
| - split: train |
| path: data/train-* |
| --- |
| |
| # Khmer Ministry Press Release Dataset (Surya-2 VLM) |
|
|
| A dataset of 99 Cambodian ministry press release document images paired with text generated via the **Surya-2 VLM**. Formatted for Vision-Language Model (VLM) fine-tuning and OCR/document-understanding tasks. |
|
|
| ## Dataset Structure |
|
|
| * `id` (`string`): Unique document identifier. |
| * `image` (`image`): Press release document scan. |
| * `messages` (`list`): Formatted chat sequence (`role`, `content`) containing prompts and extracted Khmer text. |
|
|
| ## Usage |
|
|
| ```python |
| from datasets import load_dataset |
| |
| dataset = load_dataset("your-username/dataset-name") |
| print(dataset["train"][0]) |
