Dataset · OleehyO
Images and LaTeX formulas for OCR
OleehyO/latex-formulas
Large dataset of image-LaTeX pairs for mathematical formula recognition.
What's inside
The dataset contains formula images and their corresponding LaTeX markup for image-to-text tasks. The 2025 update features nearly 80 million examples with improved quality and diversity. It was used to train the TeXTelller formula recognition model.
What you can do with it
- Train an OCR model for formula recognition
- Fine-tune a vision-language model for image-to-LaTeX
- Test domain adaptation across various formula styles
- Tasks
- Image → text
- Size
- 1M–10M rows
- Downloads this month
- 802
- Likes
- 105
Models trained on it
- OleehyO/TexTellerImage → text 29.1K
- onnx-community/TexTeller3-ONNXImage → text 357
- Ji-Ha/TexTeller3-ONNX-dynamic 196
- onnx-community/TexTeller-ONNXImage → text 167
- DGurgurov/im2latexImage → text 85
- lamm-mit/Cephalo-LaTeX-Phi-3-vision-128k-4b-alphaImage → text 38
- lamm-mit/Cephalo-Phi-3-MoE-vision-128k-3x4b-betaImage text → text 31
- Ji-Ha/TexTeller3Image → text 19
How to load it
from datasets import load_dataset
ds = load_dataset("OleehyO/latex-formulas")