This repository contains a synthetic OCR dataset generated by the Synthetic OCR Image Generator pipeline. It is intended for benchmarking and evaluating OCR or VLM systems on markdown-oriented document recognition tasks.

At a Glance

Dataset: https://huggingface.co/datasets/junyeong-nero/synthetic-ocr-images-ko Language: ko Generated samples: 1000 Requested samples: 1000 Generation mode: markdown

Included Splits

train: 900… See the full description on the dataset page: https://huggingface.co/datasets/junyeong-nero/synthetic-ocr-images-ko.