Re: Packages that work with LLMs - Moxin seems problematic too

On 8/14/26 7:32 AM, Jean Louis wrote:
> I will verify more of the LLMs which claim to be trully free, such as:
> 
> - Moxin github.com/moxin-org/Moxin-LLM -- but trained with the non-free 
> LLM?
> - OLMo (from Allen AI) allenai.org/olmo
> - Marin github.com/dlwh/marin
> - DaVinci github.com/GAIR-NLP/daVinci-LLM
> - SmolLM github.com/huggingface/smollm (good candidate)
> - OpenCoder opencoder-llm.github.io
> 
> That there are more and more efforts in getting truly free LLM is good 
> indication.
> 
> We have to teach those organization with truly free LLM would be and how 
> it should be made.
> 
> Jean Louis

Moxin seems problematic in terms of licensing as well. Their starting 
website seems to be here:

https://github.com/moxin-org/Moxin-LLM

They have this section on pretraining documentation:

https://github.com/moxin-org/Moxin-LLM/tree/main/train

Quote: "The Stack contains over 6TB of permissively-licensed source code 
files covering 358 programming languages."

This seems to be "The Stack" again, not filtered in a way where licenses 
requiring attribution requirements are somehow avoided. So I don't see 
how the output of Moxin could be adhering to the attribution desires of 
the projects found in the training data, which again doesn't seem very 
open to me.

But as before, check the according pages for yourself, and I'm still not 
a lawyer and this isn't legal advice.

Regards,

Ellie