Overview

kucingcoder/raw-ocr-to-json is a 0.5 billion parameter language model, built on the Qwen2.5 architecture and provided in GGUF format. It was fine-tuned using Unsloth, which facilitated a 2x faster training process. The model is specifically designed for converting raw Optical Character Recognition (OCR) output into structured JSON data.

Key Capabilities

• OCR to JSON Conversion: Specializes in transforming unstructured text from OCR into a parseable JSON format.
• GGUF Format: Available in GGUF format, making it compatible with various inference engines like llama-cli and llama-mtmd-cli.
• Ollama Support: Includes an Ollama Modelfile for streamlined deployment and integration.

Good For

• Automated Data Extraction: Ideal for projects requiring the extraction of structured information from scanned documents or images.
• Document Processing: Useful in workflows that involve converting document content into machine-readable JSON for further analysis or storage.
• Edge Device Deployment: The compact 0.5B parameter size and GGUF format make it suitable for deployment on devices with limited resources.