LLaMA 3.1-8B, finetuned on a synthetic OCR dataset for superior OCR correction.
Want to make some of these yourself?
