Trilogix1/Hugston-Macaron-V1-Tall

This is an optimized model of Qwen from Mind Lab then GGUFED using Quanta and HugstonOne from Hugston.

Use it with HugstonOne (the new release, Video-Audio-Image processing) , the most powerful local AI with strongest privacy worldwide to date.

Whitepaper: https://doi.org/10.5281/zenodo.21600618 ---------------- Get the latest update here: https://hugston.com

Worth mentioning additional features, Video-Audio-Image processing,-------------------Agents, skills, tools. ---------------Rag in Terabytes-------------Online mode visual interactive or background, --------Most of libraries load in preview totally offline,-------------Hugston Private API included.

The aim is to test and understand mechanism and accuracy of different llm models for research purposes.

Credit to Qwen for the model creation

Credit to https://huggingface.co/mindlab-research for the optimization method

Credit to LLama.cpp team for the great contribution

Credit to Hugston Team for Converting, Quantizing, Testing, Benching and other...

Credit to Huggingface for the amazing hosting platform

Here we show Quanta our convertor and Quantizer tool.