OCRonos is an 8 billion parameter language model developed by PleIAs, based on the Llama-3 architecture, specifically trained for the correction of badly digitized texts. It excels at rectifying OCR errors, word segmentation issues, and overall broken text structures in documents from cultural heritage and financial/administrative sources. This model is a specialized tool within the Bad Data Toolbox, designed to make challenging, deteriorated text resources usable for LLM applications and search retrieval.
No reviews yet. Be the first to review!
Hugging Face