Dataset Viewer
Auto-converted to Parquet Duplicate
image
image
pdf_stem
string
category
string
num_tests
int64
test_types
list
2502.15977_pg21
arxiv_math
1
[ "math" ]
2503.02004_pg9
arxiv_math
9
[ "math" ]
2503.037540
arxiv_math
0
[]
2503.03759_pg9
arxiv_math
4
[ "math" ]
2503.03762
arxiv_math
0
[]
2503.03765
arxiv_math
0
[]
2503.037662
arxiv_math
0
[]
2503.03772
arxiv_math
0
[]
2503.038270
arxiv_math
0
[]
2503.03847_pg30
arxiv_math
1
[ "math" ]
2503.03855_pg5
arxiv_math
9
[ "math" ]
2503.03861_pg30
arxiv_math
3
[ "math" ]
2503.03873_pg5
arxiv_math
6
[ "math" ]
2503.03879_pg4
arxiv_math
8
[ "math" ]
2503.03899_pg9
arxiv_math
7
[ "math" ]
2503.03903_pg9
arxiv_math
3
[ "math" ]
2503.03905_pg7
arxiv_math
10
[ "math" ]
2503.039094
arxiv_math
0
[]
2503.03948_pg3
arxiv_math
4
[ "math" ]
2503.03949
arxiv_math
0
[]
2503.03952_pg5
arxiv_math
4
[ "math" ]
2503.0399408
arxiv_math
0
[]
2503.04024_pg4
arxiv_math
4
[ "math" ]
2503.04026_pg2
arxiv_math
2
[ "math" ]
2503.04033_pg9
arxiv_math
5
[ "math" ]
2503.04040_pg2
arxiv_math
10
[ "math" ]
2503.04041_pg20
arxiv_math
3
[ "math" ]
2503.04045_pg2
arxiv_math
1
[ "math" ]
2503.04047_pg8
arxiv_math
3
[ "math" ]
2503.04048_pg46
arxiv_math
1
[ "math" ]
2503.04056_pg25
arxiv_math
1
[ "math" ]
2503.04068_pg2
arxiv_math
1
[ "math" ]
2503.04077
arxiv_math
0
[]
2503.04086_pg3
arxiv_math
2
[ "math" ]
2503.040920
arxiv_math
0
[]
2503.04108_pg45
arxiv_math
2
[ "math" ]
2503.04116_pg9
arxiv_math
2
[ "math" ]
2503.041244
arxiv_math
0
[]
2503.04147_pg3
arxiv_math
10
[ "math" ]
2503.04182_pg3
arxiv_math
8
[ "math" ]
2503.041891
arxiv_math
0
[]
2503.04226_pg3
arxiv_math
10
[ "math" ]
2503.04228_pg5
arxiv_math
7
[ "math" ]
2503.04238_pg26
arxiv_math
6
[ "math" ]
2503.04245_pg7
arxiv_math
5
[ "math" ]
2503.04247_pg22
arxiv_math
2
[ "math" ]
2503.042510
arxiv_math
0
[]
2503.04255
arxiv_math
0
[]
2503.04297_pg20
arxiv_math
2
[ "math" ]
2503.04323_pg24
arxiv_math
6
[ "math" ]
2503.04329_pg7
arxiv_math
10
[ "math" ]
2503.04382_pg8
arxiv_math
7
[ "math" ]
2503.04397_pg2
arxiv_math
6
[ "math" ]
2503.04407_pg3
arxiv_math
10
[ "math" ]
2503.04415_pg2
arxiv_math
2
[ "math" ]
2503.04425_pg8
arxiv_math
10
[ "math" ]
2503.04430_pg9
arxiv_math
9
[ "math" ]
2503.044330
arxiv_math
0
[]
2503.044386
arxiv_math
0
[]
2503.044480
arxiv_math
0
[]
2503.04465_pg3
arxiv_math
4
[ "math" ]
2503.04466_pg9
arxiv_math
10
[ "math" ]
2503.04467_pg7
arxiv_math
9
[ "math" ]
2503.04471_pg7
arxiv_math
2
[ "math" ]
2503.04486_pg21
arxiv_math
3
[ "math" ]
2503.04488_pg9
arxiv_math
10
[ "math" ]
2503.04493_pg31
arxiv_math
10
[ "math" ]
2503.04494_pg21
arxiv_math
1
[ "math" ]
2503.04498_pg27
arxiv_math
1
[ "math" ]
2503.04523_pg4
arxiv_math
1
[ "math" ]
2503.045279
arxiv_math
0
[]
2503.04535_pg2
arxiv_math
4
[ "math" ]
2503.04536_pg9
arxiv_math
10
[ "math" ]
2503.04555_pg2
arxiv_math
2
[ "math" ]
2503.04567_pg38
arxiv_math
1
[ "math" ]
2503.04577_pg2
arxiv_math
1
[ "math" ]
2503.04578_pg2
arxiv_math
6
[ "math" ]
2503.04583_pg3
arxiv_math
1
[ "math" ]
2503.04590_pg7
arxiv_math
4
[ "math" ]
2503.04604_pg3
arxiv_math
7
[ "math" ]
2503.04607_pg2
arxiv_math
1
[ "math" ]
2503.04612_pg8
arxiv_math
10
[ "math" ]
2503.04620_pg35
arxiv_math
7
[ "math" ]
2503.04623_pg54
arxiv_math
4
[ "math" ]
2503.04646_pg2
arxiv_math
3
[ "math" ]
2503.046494
arxiv_math
0
[]
2503.04667_pg3
arxiv_math
6
[ "math" ]
2503.046749
arxiv_math
0
[]
2503.04678_pg20
arxiv_math
2
[ "math" ]
2503.04690_pg4
arxiv_math
1
[ "math" ]
2503.04701_pg26
arxiv_math
3
[ "math" ]
2503.04876
arxiv_math
0
[]
2503.04881_pg25
arxiv_math
2
[ "math" ]
2503.04897_pg4
arxiv_math
10
[ "math" ]
2503.04912_pg4
arxiv_math
1
[ "math" ]
2503.04917
arxiv_math
0
[]
2503.04923_pg9
arxiv_math
10
[ "math" ]
2503.04932_pg20
arxiv_math
4
[ "math" ]
2503.04950_pg23
arxiv_math
10
[ "math" ]
2503.04958_pg4
arxiv_math
6
[ "math" ]
End of preview. Expand in Data Studio

olmOCR-bench Pre-Rendered

Pre-rendered PNG images of the olmOCR-bench benchmark dataset, ready for zero-setup evaluation of any OCR / vision model.

What This Is

The official olmOCR benchmark requires downloading 1,403 PDFs locally and rendering each page to a PNG image before sending it to a model. Every benchmark runner in the official repo does this same rendering step internally — see olmocr/data/renderpdf.py::render_pdf_to_base64png().

This dataset eliminates that setup entirely by hosting the pre-rendered images directly. The PNGs are rendered at target_longest_image_dim=2048 — the same default resolution used by the official olmOCR render_pdf_to_base64png() function and by the GPT-4o, Claude, and Gemini benchmark runners.

All files are accessible via direct URL, so you can evaluate any model by just pointing at these URLs — no local downloads, no PDF rendering tools, no dataset cloning.

Dataset Structure

The dataset has 7 subsets (one per benchmark category), each with a test split:

Subset PDFs Tests Test Types
arxiv_math 522 2,927 math
headers_footers 266 753 absent
long_tiny_text 62 442 present
multi_column 231 884 order
old_scans 98 526 present, absent, order
old_scans_math 36 458 math
tables 188 1,020 table

Loading with datasets

from datasets import load_dataset

ds = load_dataset("shhdwi/olmocr-pre-rendered", "arxiv_math", split="test")
print(ds[0])  # {'image': <PIL.Image>, 'pdf_stem': '...', 'category': '...', ...}

Contents

Directory Contents Count
images/ Pre-rendered PNG images (page 1, 2048px longest dim) with metadata.jsonl per category 1,403 images
ground_truth/ JSONL test case files (from allenai/olmOCR-bench) 7,010 tests
predictions/ Published model prediction caches 1,403 .md files

Rendering Details

Each PDF page is rendered to PNG matching the official olmOCR benchmark process:

  • Resolution: target_longest_image_dim = 2048 (longest side scaled to 2048px, aspect ratio preserved)
  • Renderer: PyMuPDF (same pixel output as pdftoppm used in the official repo)
  • Pages: Page 1 only (the benchmark tests only page 1 of each PDF)
  • Naming: {pdf_stem}_pg1.png

This matches what the official benchmark runners do internally:

  • run_chatgpt.py: render_pdf_to_base64png(pdf_path, target_longest_image_dim=2048)
  • run_claude.py: render_pdf_to_base64png(pdf_path, target_longest_image_dim=2048)
  • run_gemini.py: render_pdf_to_base64png(pdf_path, target_longest_image_dim=2048)
  • run_server.py: render_pdf_to_base64png(pdf_path, target_longest_image_dim=1024) (for smaller models)

Quick Start

Evaluate any model with zero setup:

pip install litellm httpx

# Run on any litellm-supported model
python run_bench.py --model gpt-4o
python run_bench.py --model claude-sonnet-4-20250514
python run_bench.py --model gemini/gemini-2.0-flash

# Run specific categories only
python run_bench.py --model gpt-4o --categories arxiv_math headers_footers

# Evaluate published predictions (no API key needed)
python run_bench.py --evaluate nanonets-optimal-v4

Direct File Access

Every file is accessible via URL:

https://huggingface.co/datasets/shhdwi/olmocr-pre-rendered/resolve/main/images/arxiv_math/2503.05390_pg14_pg1.png
https://huggingface.co/datasets/shhdwi/olmocr-pre-rendered/resolve/main/ground_truth/arxiv_math.jsonl

Attribution

Based on olmOCR-bench by Allen AI (paper). Licensed ODC-BY-1.0.

Downloads last month
51

Paper for shhdwi/olmocr-pre-rendered