Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringclasses
2 values
side
stringclasses
2 values
image
imagewidth (px)
576
576
markdown
stringclasses
2 values
inference_info
stringclasses
1 value
permis-recto
recto
<div style="text-align: right; margin-top: -10px; margin-right: 20px;"> <div style="font-size: 10px; line-height: 1.2;"> 55 344 3122<br> CAN 373473 </div> </div> <div style="text-align: center; margin-top: 200px; margin-bottom: 200px;"> <table border="1" cellpadding="5" cellspacing="0" styl...
[{"model_id": "lightonai/LightOnOCR-2-1B", "model_name": "LightOnOCR-2", "column_name": "markdown", "timestamp": "2026-08-11T13:50:53.497995", "temperature": 0.2, "top_p": 0.9, "max_tokens": 4096, "target_size": 1540}]
permis-verso
verso
<table border="1" class="dataframe"> <thead> <tr style="text-align: right;"> <th>الأقسام</th> <th>تاريخ التعلم</th> <th>المقيّمات</th> </tr> <tr> <th>Catégories</th> <th>Date de délivrance</th> <th>Restrictions</th> </tr> </thead> <tbody> <tr> <td>A1</...
[{"model_id": "lightonai/LightOnOCR-2-1B", "model_name": "LightOnOCR-2", "column_name": "markdown", "timestamp": "2026-08-11T13:50:53.497995", "temperature": 0.2, "top_p": 0.9, "max_tokens": 4096, "target_size": 1540}]

Document OCR using LightOnOCR-2-1B

This dataset contains OCR results from images in mehdigououiad/permis-ocr-bench using LightOnOCR-2, a fast and compact 1B OCR model trained with RLVR.

Processing Details

Configuration

  • Image Column: image
  • Output Column: markdown
  • Dataset Split: train
  • Batch Size: 16
  • Target Image Size: 1540px (longest dimension)
  • Max Model Length: 16,384 tokens
  • Max Output Tokens: 4,096
  • Temperature: 0.2
  • Top P: 0.9
  • GPU Memory Utilization: 80.0%

Model Information

LightOnOCR-2 is a next-generation fast, compact OCR model that excels at:

  • Fastest Speed - 42.8 pages/second on H100 GPU (7× faster than v1)
  • 🎯 High Accuracy - 83.2 ± 0.9% on OlmOCR-Bench (+7.1% vs v1)
  • 🧠 RLVR Training - Eliminates repetition loops and formatting errors
  • 📚 Better Dataset - 2.5× larger training data with cleaner annotations
  • 📐 LaTeX formulas - Mathematical notation in LaTeX format
  • 📊 Tables - Extracted and formatted as markdown
  • 📝 Document structure - Hierarchy and layout preservation
  • 🌍 Multilingual - Optimized for European languages
  • 💪 Production-ready - Outperforms models 9× larger

Key Improvements over v1

  • 7.5× faster: 42.8 vs 5.71 pages/sec on H100
  • +7.1% accuracy: 83.2% vs 76.1% on benchmarks
  • Better quality: RLVR training eliminates common OCR errors
  • Cleaner output: No repetition loops or formatting glitches
  • Simpler: Single model (no vocabulary variants)

Dataset Structure

The dataset contains all original columns plus:

  • markdown: The extracted text in markdown format with LaTeX formulas
  • inference_info: JSON list tracking all OCR models applied to this dataset

Usage

from datasets import load_dataset
import json

# Load the dataset
dataset = load_dataset("{output_dataset_id}", split="train")

# Access the markdown text
for example in dataset:
    print(example["markdown"])
    break

# View all OCR models applied to this dataset
inference_info = json.loads(dataset[0]["inference_info"])
for info in inference_info:
    print(f"Column: {info['column_name']} - Model: {info['model_id']}")

Reproduction

This dataset was generated using the uv-scripts/ocr LightOnOCR-2 script:

uv run https://huggingface.co/datasets/uv-scripts/ocr/raw/main/lighton-ocr2.py \
    mehdigououiad/permis-ocr-bench \
    <output-dataset> \
    --image-column image \
    --batch-size 16

Performance

  • Processing Speed: ~0.02 images/second
  • Benchmark Score: 83.2 ± 0.9% on OlmOCR-Bench
  • Training: RLVR (Reinforcement Learning with Verifiable Rewards)

Generated with 🤖 UV Scripts

Downloads last month
-