Meta releases big, new open-source AI big language model

Meta, which most of us know as Facebook, has Commercial version of Llama-v2 releasedIts open-source Large Language Model (LLM) that uses artificial intelligence (AI) to generate text, images and code.

The first version of the Large Language Model Meta AI (LAMA) was publicly announced in February and was restricted to approved researchers and organizations. However, it was soon leaked online In early March for anyone to download and use.

Meta coincidentally filed take-down orders to restrain stolen programs on sites such as GitHub and the open-source AI group Hugging Face. Ultimately, faced with the code being readily available on the web, Meta gave up trying to order Tidal back. Instead, it embraced release.

Too: Do you want to build your own AI Chatbot? Say hello to open-source HuggingChat

Both versions of Llama are trained on CommonCrawl, GitHub, Wikipedia, Project Gutenberg, arXiv, Stack Exchange, and other open test websites. While Microsoft and OpenAI’s ChatGPT grabbed the headlines, many open-source developers turned to Llama.

In addition to access to the llama, the meta also shared its weight. Other major LL.M.’s have not done this. With weights, the parameters learned by a model during training, it is very easy to build and run custom AI programs. Other large LLMs, such as the GPT, are generally only accessible through application programming interfaces (APIs).

Whereas AI is built on an open-source foundation, LLAMA is the first major open-source LL.M. Its pre-trained model is trained on 2 trillion tokens, and it has to double the context length of llama1. Its fine-grained models have been trained on over 1 million human annotations. Its model size parameters range from 7 to 70 billion parameters.

Too: Microsoft announces Azure AI Trio at Inspire 2023

So, that’s why open-source developers welcomed Llama, but Why did meta open source Llama-v2?, According to its researchers, “While many companies choose to build AI behind closed doors, we are releasing Llama 2 openly to encourage responsible AI innovation. Based on our experience, an open approach is based on collective knowledge, diversity, and ingenuity. The AI ​​practitioner community will realize the benefits of this technology. Collaboration will make these models better and safer.”

But, is Llama-v2 really open source? While Meta says the right things, Llama 2 Community License Agreement Not approved by the Open Source Initiative (OSI), the gold-standard group for open-source licensing.

With this in mind, Meta has created Llama 2 is available free for research and commercial use, The meta also includes model weights and initialization code for pre-trained models and convolutional fine-tuned versions. This marks a huge step forward for developers to bring Llama-powered applications to use.

Too: 5 Ways to Explore the Use of Generative AI at Workplace

Strictly open source or not, Llama 2 is certainly open enough for most practical purposes. and, Amanda Brock, as CEO OpenUKto put it, it’s not an OSI approved license but a Important release of open technology …This is a step towards taking AI out of the hands of the few and into the hands of the many, democratizing technology and its use through transparency and building trust in the future. No, it’s not perfect, and yes, there is more work to be done, but this bold step towards AI-open innovation with a responsible but light-hearted principle-based approach to regulating AI use and development sets the path.

Meta just hasn’t been playing nice with the open-source community. META also announced that Microsoft is its preferred Llama 2 partner. So, Llama 2 is available in the Azure AI Model Catalog, which enables developers using Microsoft Azure to build with it. LMA is also optimized to run natively on Windows, giving developers a seamless workflow as they bring Generative AI experiences to customers across a variety of platforms. Llama 2 is available through Amazon Web Services (AWS), Hugging Faces, and other providers.

Too: Microsoft’s Inspire 2023: 10 AI and partnership announcements to know

In addition to opening up the code, in this release, Meta is also attempting to make the Llama safer, more well-behaved, and less prone to hallucinations than other models.

META does this by red-teaming (security testing) exercises designed to address security vulnerabilities. Meta has also issued Developer’s guide to safe and responsible usage Helping developers understand and apply best practices for developing and responsible model testing. Ultimately, Meta provided a llama. Acceptable Use Policy Prohibiting certain use cases to help ensure that these models are being used fairly and responsibly.

Final result? Meta hopes to catch up and surpass OpenAI. Who knows, it might be able to do this. As a Google AI engineer recently wrote, “The inconvenient truth is, We are not in a position to win this (Generative AI) arms race, and neither is OpenAI, While we were fighting, a third group was quietly eating our lunch.” That third group? open source community,