lightonocr

This 2.7GB Local OCR Model Changed My Workflow! (LightOnOCR-3 Review) - High-accuracy offline document understanding pipeline with Ollama and GGUF.

Mistral-OCR-compatible document OCR service for LibreChat on top of vLLM and LightOnOCR

High-performance OCR microservice powered by LightOnOCR-2-1B. Handles complex documents (tables, math) with dual inference engines (Hugging Face & vLLM) and a production-ready Docker setup.

LoRA for LightOnOCR-3-4B that fixes math-notation OCR errors (primes, Greek letters, bold/bar, I vs 1), with a CC BY/CC0 equation dataset, a frozen benchmark and a rerunnable scorer

A minimal inference client, CLI, and interactive web viewer for LightOnOCR (LightOnOCR 1, 2, and 3) served via vLLM. This tool provides side-by-side document layout inspection, bounding-box grounding overlays, and clean Markdown transcription rendering.