Meta Launches Llama 2: Open Source and Free for Commercial Use
Meta has released Llama 2, a new generation of the company's large language model, which is now open source and available for free for both commercial use and research. The launch is conducted in partnership with Microsoft.
Vad har hänt
On 19 July 2023, Meta launched Llama 2, an updated version of its large-scale language model. Unlike its predecessor LLaMA, which was intended for research purposes, Llama 2 is freely available for commercial use and research. The launch was announced during Microsoft's Inspire Partner Conference, marking a deeper collaboration between Meta and Microsoft.
Key facts
| Lanseringsdatum | 19 juli 2023 |
|---|---|
| Modellstatus | Öppen källkod, gratis |
| Kommersiell användning | Tillåtet |
| Partnerskap | Microsoft |
| Antal parametrar | 7B, 13B, 70B |
| Träningsdata ökning | 40% |
”Meta har en lång historia av öppen källkod, vilket driver innovation. Det gör det möjligt för fler utvecklare att bygga med ny teknik och förbättrar också teknikens säkerhet. Vi tror att om ekosystemet är mer öppet, kommer fler framsteg att ske, och det är därför vi har gjort Lla”
Varför det spelar roll
By making Llama 2 open source and permitting commercial use, Meta is challenging the more closed strategies of competitors like OpenAI's GPT-4 and Google's PaLM 2. Meta CEO Mark Zuckerberg has stated that open source fosters innovation, increases accessibility for developers, and improves security. This move aims to create a more open ecosystem for the development of large language models.
Vem påverkas
Developers, researchers, and companies looking to build applications based on large language models are directly affected, as they now have access to a powerful model without licensing costs for commercial use. Meta and Microsoft also benefit from the expanded distribution and potential for ecosystem building. Competitors like OpenAI and Google face a new open-source strategy in the market for large language models.
EU-status
Llama 2 is globally available, meaning users and developers within the EU can benefit from the model. The open-source licence supports broad use outside the United States.
Mer att veta
Llama 2 represents a significant upgrade over the previous LLaMA model, featuring 40% more training data and models in sizes of 7 billion, 13 billion, and 70 billion parameters. It has been trained on 2 trillion tokens and possesses double the context length of its predecessor.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vem påverkas?
Påverkar det EU?
Länken öppnar i nytt fönster och leder till utgivarens egen sida.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.