---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: ground_truth
    dtype: string
  - name: labels
    dtype: string
  splits:
  - name: train
    num_bytes: 339139900
    num_examples: 4000
  download_size: 332704349
  dataset_size: 339139900
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---


deepseek ocr labeled