NuExtract 1.5
About
NuExtract 1.5 is an advanced multilingual, open-source foundation model developed by NuMind, designed specifically for structured information extraction. It is a fine-tuned version of the Phi-3.5-mini-instruct model, trained on a premium private dataset to enhance its capabilities. Unlike standard language models that primarily focus on text generation, NuExtract 1.5 excels at extracting and returning structured data in JSON format directly from input text. One of its standout features is handling a vast range of document lengths, outperforming larger models like GPT-4o in English language tasks while maintaining a significantly smaller footprint. Additionally, this model supports multiple languages including English, French, Spanish, German, Portuguese, and Italian, making it a versatile tool for a variety of applications 14.