Datasets:

davanstrien
/

sample-firered-ocr

image imagewidth (px) 596 596	raw_queries listlengths 1 1	broad_topical_query stringclasses 2 values	broad_topical_explanation stringclasses 2 values	specific_detail_query stringclasses 2 values	specific_detail_explanation stringclasses 2 values	visual_element_query stringclasses 2 values	visual_element_explanation stringclasses 2 values	parsed_into_json bool 1 class	markdown stringclasses 2 values	inference_info stringclasses 1 value
	[ "{\n \"broad_topical_query\": \"1954 Blue Book Project\",\n \"broad_topical_explanation\": \"This query targets the specific year and project mentioned in the document, which is likely to be a unique identifier for this document.\",\n \"specific_detail_query\": \"L'astronomie et la lune\",\n \"specific_detail_e...	1954 Blue Book Project	This query targets the specific year and project mentioned in the document, which is likely to be a unique identifier for this document.	L'astronomie et la lune	This query focuses on the specific topic of astronomy and the moon, which is a central theme in the document.	Page 0	This query references the page number, which is a unique identifier for this document within a larger corpus.	true	AVRIL 1954 Projet "Blue Book" TOP SECRET Pourquoi les astronautes ne veulent pas parler de la lune Des villes qui croissent,des signaux lumineux,des explosions et des ombres géométriques étranges-ooci et d'autres phénomènes inexpli cables par des lois naturelles connues ont maintenent été observés sur la lune par des ...	[{"model_id": "FireRedTeam/FireRed-OCR", "column_name": "markdown", "timestamp": "2026-06-05T10:30:10.187906"}]
	[ "{\n \"broad_topical_query\": \"Document about the study of Alhazen's work on optics\",\n \"broad_topical_explanation\": \"This query focuses on the main subject of the document, which is the study of Alhazen's work on optics.\",\n \"specific_detail_query\": \"Alhazen's work on optics\",\n \"specific_detail_exp...	Document about the study of Alhazen's work on optics	This query focuses on the main subject of the document, which is the study of Alhazen's work on optics.	Alhazen's work on optics	This query targets a specific detail mentioned in the document, which is Alhazen's work on optics.	Document with a reference to Alhazen's work on optics	This query looks for a document that contains a reference to Alhazen's work on optics, which is a visual element in the document.	true	10 Au début du siècle dernier l'astronomie Schroeter étudie les oépric- ces de la lune per l'évenons et le retrait du bord ensolliée. Comme point de repère pour la mensuration, il choisit le crétère Alhazen larges de 23 milles dans Mare Crisium, per os qu'il était près du bord nord-est et était très visible sous forme...	[{"model_id": "FireRedTeam/FireRed-OCR", "column_name": "markdown", "timestamp": "2026-06-05T10:30:10.187906"}]

Document OCR using FireRed-OCR

This dataset contains OCR results from images in davanstrien/ufo-ColPali using FireRed-OCR, a 2.1B model fine-tuned from Qwen3-VL-2B-Instruct.

Processing Details

Source Dataset: davanstrien/ufo-ColPali
Model: FireRedTeam/FireRed-OCR
Number of Samples: 2
Processing Time: 2.3 min
Processing Date: 2026-06-05 10:30 UTC

Configuration

Image Column: image
Output Column: markdown
Dataset Split: train
Batch Size: 16
Max Model Length: 32,768 tokens
Max Output Tokens: 8,192
GPU Memory Utilization: 80.0%

Model Information

FireRed-OCR is a document OCR model that converts images to structured Markdown:

Fine-tuned from Qwen3-VL-2B-Instruct (2.1B parameters)
LaTeX formula support (inline and block)
HTML table extraction
Layout-aware text extraction
Apache 2.0 licensed

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 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 FireRed-OCR script:

uv run https://huggingface.co/datasets/uv-scripts/ocr/raw/main/firered-ocr.py \
    davanstrien/ufo-ColPali \
    <output-dataset> \
    --image-column image \
    --batch-size 16 \
    --max-model-len 32768 \
    --max-tokens 8192 \
    --gpu-memory-utilization 0.8

Generated with UV Scripts

Downloads last month: 11

Number of rows:

Total file size:

282 kB