README.md

metadata

tags:
  - ocr
  - document-processing
  - qianfan-ocr
  - markdown
  - uv-script
  - generated

Document OCR using Qianfan-OCR

This dataset contains OCR results from biglam/rubenstein-manuscript-catalog using Qianfan-OCR, Baidu's 4.7B end-to-end document intelligence model.

Processing Details

• Source Dataset: biglam/rubenstein-manuscript-catalog
• Model: baidu/Qianfan-OCR
• Number of Samples: 5
• Processing Time: 6.6 min
• Processing Date: 2026-03-19 15:50 UTC

Configuration

• Image Column: image
• Output Column: markdown
• Dataset Split: train
• Batch Size: 8
• Prompt Mode: ocr
• Layout-as-Thought: Enabled
• Thinking Traces: Excluded
• Max Model Length: 16,384 tokens
• Max Output Tokens: 8,192
• GPU Memory Utilization: 85.0%

Model Information

Qianfan-OCR key capabilities:

• #1 end-to-end model on OmniDocBench v1.5 (93.12)
• #1 on OlmOCR Bench (79.8)
• 192 language support
• Layout-as-Thought reasoning for complex documents
• Document parsing, table extraction, formula recognition, chart understanding
• Key information extraction

Dataset Structure

The dataset contains all original columns plus:

• markdown: The extracted text in markdown format
• inference_info: JSON list tracking all OCR models applied

Reproduction

uv run https://huggingface.co/datasets/uv-scripts/ocr/raw/main/qianfan-ocr.py \
    biglam/rubenstein-manuscript-catalog \
    <output-dataset> \
    --image-column image \
    --prompt-mode ocr \
    --batch-size 8  --think

Generated with UV Scripts