Meta's LLaMa 2 license is not an open source license

fc36a36a63a74f7e59c6bd4c4d059141.jpeg

0ea0efc2e4ce6d7037e45ca5b9e8d125.jpeg

The author was honored to be invited to participate in the Open Source Congress held by the Linux Foundation in Geneva, Switzerland on July 27. The agenda is as follows:

https://oscongress2023.sched.com/。

I saw some reports and articles recommended for participants to read in advance during the interactive discussion session. Among them was an OSI blog "Meta's LLaMa 2 license is not an open source license", which particularly caught my attention, and thus obtained the OSI agreed to translate it for the reference of domestic audiences.

       Liu Tiandong

Co-founder and official member of Kaiyuanshe

OSI is pleased to see that Meta is lowering the barrier to entry for powerful AI systems. Unfortunately, the tech giant has given the false impression that LLaMa 2 is "open source" when it is not. Even assuming that the term can be effectively applied to large language models composed of multiple resources of different types, Meta confuses "open source" with "resources made available to certain users under certain conditions", which are two entirely different things . We have asked them to correct misstatements.

OSI is pleased to see that Meta is lowering barriers for access to powerful AI systems. Unfortunately, the tech giant has created the misunderstanding that LLaMa 2 is “open source” – it is not. Even assuming the term can be validly applied to a large language model comprising several resources of different kinds, Meta is confusing “open source” with “resources available to some users under some conditions,” two very different things. We’ve ask them to correct their misstatement.

"Open Source" means software under a license with certain characteristics as defined in the Open Source Definition (OSD). Among other requirements, the license does not discriminate against individuals, groups or areas of work (OSD points 5 and 6). Meta's LLaMa model and code license does not meet this standard; specifically, it places restrictions on commercial use by certain users (section 2), and also restricts the use of models and software for certain purposes ("may Accept the Use Policy").

“Open Source” means software under a license with specific characteristics, defined by the Open Source Definition (OSD). Among other requirements, for a license to be Open Source, it may not discriminate against persons or groups or fields of endeavor (OSD points 5 and 6). Meta’s license for the LLaMa models and code does not meet this standard; specifically, it puts restrictions on commercial use for some users (paragraph 2) and also restricts the use of the model and software for certain purposes (the Acceptable Use Policy). 

bca4c0f27057faa17cf2de4361ff8799.jpeg

Open source licenses ensure that developers and users can decide for themselves how and where to use technology without working with another party; they have sovereignty over the technology they use. The premise of open source is that everyone can share, no matter who you are. The commercial restrictions in Section 2 of the LLAMA Community License Agreement run counter to the promises in the OSD.

An Open Source license ensures that developers and users are able to decide for themselves how and where to use the technology without the need to engage with another party; they have sovereignty over the technology they use. Open Source is premised on the understanding that everyone gets to share no matter who you are. The commercial limitation in paragraph 2 of LLAMA COMMUNITY LICENSE AGREEMENT is contrary to that promise in the OSD.

OSI doesn't question Meta's desire to limit the use of Llama for competitive purposes, but doing so would take the license out of the "open source" category.

OSI does not question Meta’s desire to limit the use of Llama for competitive purposes, but doing so takes the license out of the category of “Open Source.”  

OSD doesn't allow to limit the field of use , because you can't know in advance what will happen in the future, good or bad. That's why the Linux kernel is popular in medical devices, airplanes, and rockets.

The OSD does not allow restrictions on field of use because you can’t know beforehand what can happen in the future, good or bad. That’s what allows the Linux kernel to become popular in medical devices as well as airplanes and rockets.

However, Meta policy prohibits its use in some areas where it may be of great benefit to society, such as the use of controlled/controlled substances and critical infrastructure. Even a simple-sounding rule like "you must obey the law" can be problematic in practice. What if the laws in different places are inconsistent? What if the law is unjust?

But the Meta policy prohibits use in several areas that might be highly beneficial to society, such as regulated/controlled substances and use for critical infrastructure. Even something that sounds as simple as “you must follow the law” is problematic in practice.  What if the law in different places is inconsistent? What if the law is unjust?

4046c27c1498ff3b5af9089b091db761.jpeg

Llama LLM's license is clearly not an "open source" license. Meta makes some aspects of its large language models available to some, but not everyone, and not for any purpose. OSI recognizes how important it is to reach a consensus on what openness means for AI systems. These are new creations of man, just as software was a new creation of the human intellect in the 1970s. We're hosting a series of events to develop a common definition of "open" in AI, and we welcome your ideas.

The license for the Llama LLM is very plainly not an “Open Source” license. Meta is making some aspect of its large language model available to some, but not to everyone, and not for any purpose. OSI realizes how important it is to come to a shared understanding of what open means for AI systems. These are new human artifacts, much like software was a new creation of human intellect in the 70s. We’re running a series of events to craft a common definition of “open” in the AI context and we welcome submissions of ideas.

Reprinted from丨Open Source Rainforest

Editor丨Weng Peipei

Related Reading| Related Reading

7ce093cf553cde6a160d39fcd028e9af.jpeg

coming! On July 30, the establishment ceremony of KCC Hangzhou Station and the first offline reading club activity invites you to participate!

b15692d14dd055efd4f3d20c648ab226.jpeg

KCC@Singapore helps AI&GPT application innovation Hackathon, win $26,000 in the last 4 days!

Introduction to Kaiyuanshe

Founded in 2014, Kaiyuan Society is composed of individual members who voluntarily contribute to the cause of open source. It is formed according to the principle of "contribution, consensus, and co-governance". It has always maintained the characteristics of vendor neutrality, public welfare, and non-profit. International integration, community development, project incubation" is an open source community federation with the mission. Kaiyuanshe actively cooperates closely with communities, enterprises and government-related units that support open source. With the vision of "Based in China and Contributing to the World", it aims to create a healthy and sustainable open source ecosystem and promote China's open source community to become an active force in the global open source system. Participation and Contributors.

In 2017, Kaiyuanshe was transformed into an organization composed entirely of individual members, operating with reference to the governance model of top international open source foundations such as ASF. In the past nine years, it has connected tens of thousands of open source people, gathered thousands of community members and volunteers, hundreds of lecturers at home and abroad, and cooperated with hundreds of sponsors, media, and community partners.

6b6e7bfd0f57181899708e8e0625778c.gif

Guess you like

Origin blog.csdn.net/kaiyuanshe/article/details/131989864