Why allenai deceptfully claims this is "Open Source"? It is not!
Tülu3 is a leading instruction following model family, offering fully open-source data, code, and recipes designed to serve as a comprehensive guide for modern post-training techniques. Tülu3 is designed for state-of-the-art performance on a diversity of tasks in addition to chat, such as MATH, GSM8K, and IFEval.
Meta’s LLaMa 2 license is not Open Source – Open Source Initiative:
https://opensource.org/blog/metas-llama-2-license-is-not-open-source
The Open Source Definition – Open Source Initiative:
https://opensource.org/osd
What is Free Software? - GNU Project - Free Software Foundation:
https://www.gnu.org/philosophy/free-sw.html
Your organization knows very well what is Open Source and what is not. You intentionally and deceptively deceive people with the purpose to gain media attention and enter into community of free software users.
You claim to compete with DeepSeek, but you publish model which is in different category of proprietary software.
If you would be using right terminology, and not claiming how it is Open Source, I wouldn't say anything. But you do.
You wish to gain followers by telling how it is all Open Source, while it is not. META didn't make that LLAMA Open Source, and so this is also not Open Source and cannot compete with DeepSeek in same category.
Stop deceiving public. Personally I feel deeply betrayed.
It's a mess. Regular EULA's are sometimes less restrictive than this.
- Meta Llama 3.1 Community License Agreement
- Tülu3's "Responsible Use Guidelines"
- Gemma's ToS
- Qwen License Agreement
- "Use it for research and education only please"
Honestly, hurtful to the industry.
Hey @JLouisBiz and @Araki we are sorry about the mix up in communication. We agree that there's nothing else to say but being better as communicating as a team. Trying to land the message of "fully open post-training tools." We'll keep working as a team and community to be better about it!
For these model cards, given they're used in many places, I'm working on a fix now to better communicate the difference between post training and broader open source lms.
It seems AllenAI only followed the path of META abusing Open Source definition.