It is quite amazing that now it is possible to run amazing the latest advanced models but,
we have however established by throughout testing that the older versions are to a better accuracy and can process AI generated or modified images.
It must be specific version to work well with VL models. We will keep updated the website with all the versions that work error free.
Big thanks to especially Qwen, team and all the teams that contributed to open source/weights for their amazing work (they never stop 24/7, and Ggerganov: https://huggingface.co/ggml-org and all the hardworking team behind llama.cpp.
Also big thanks to Huggingface.co team for their incredible contribution.
Lastly Thank you to the Hugston Team that never gave up and made all this possible.
Enjoy
PS: we are on the way to a bug free error Qwen3 80B GGUF
trilogic•3mo ago
in HugstonOne Enterprise Edition 1.0.8 (Available here: https://hugston.com/uploads/software/HugstonOne%20Enterprise...
Now they work quite good.
We noticed that every version has a bug:
1- They do not process the AI Images
2 They do not process the Modified Images.
It is quite amazing that now it is possible to run amazing the latest advanced models but, we have however established by throughout testing that the older versions are to a better accuracy and can process AI generated or modified images.
It must be specific version to work well with VL models. We will keep updated the website with all the versions that work error free.
Big thanks to especially Qwen, team and all the teams that contributed to open source/weights for their amazing work (they never stop 24/7, and Ggerganov: https://huggingface.co/ggml-org and all the hardworking team behind llama.cpp.
Also big thanks to Huggingface.co team for their incredible contribution.
Lastly Thank you to the Hugston Team that never gave up and made all this possible.
Enjoy
PS: we are on the way to a bug free error Qwen3 80B GGUF