Social media giant Meta, previously known as Facebook, has announced the launch of its new artificial intelligence (AI) system - Llama 2. The system, described as the next generation of its AI large language model, is intended to rival OpenAI’s ChatGPT and Google's Bard. The noteworthy aspect of this technology is that it will be open-sourced and free for both research and commercial use.

Llama 2 has been designed to empower developers and organizations with the ability to build AI-powered tools and experiences. In a blog post, the company expressed their astonishment at the high demand for the first version of Llama, with over 100,000 requests for access from researchers. The learnings and success of Llama 1 have contributed to the development of Llama 2, which is now open-sourced and available free of charge.

According to Meta, Llama 2 models have been trained on two trillion tokens, twice the context length of the predecessor. These models have also been trained on over a million new human annotations. In terms of availability and accessibility, Azure customers can fine-tune and deploy the 7B, 13B, and 70B-parameter Llama 2 models on the platform with ease and safety.

While the company's recent research paper introducing the new model has been less open than previous papers, Meta affirms that the model was trained on a new mix of publicly available data, excluding any data from Meta’s products or services. As a measure of privacy, the company claims to have removed data from websites known to contain a high volume of personal information about private individuals.

Llama 2 can be accessed through the Azure AI model catalog, offering Microsoft Azure users an opportunity to build with it and leverage their cloud-native tools for content filtering and safety features. It is also optimized to run locally on Windows, thus giving developers a seamless workflow. Apart from Microsoft Azure, Llama 2 is available through Amazon Web Services (AWS), Hugging Face, and other providers.

When compared to ChatGPT and Bard, Llama 2 not only outperforms other open-source language models on many external benchmarks but also offers open-source access for research and commercial use. Llama 2 has been trained on 40% more data than its predecessor, with over a million human annotations for fine-tuning the quality of outputs.

As a part of its strategy, Meta emphasizes the benefits of being more open than its rivals in the Big Tech industry, making it easier for external researchers to identify and mitigate the bias and toxicity in AI systems. The open-source nature of Llama 2 is perceived as a significant driver for innovation, enhancing safety and security through collective scrutiny and problem-solving.

Lastly, Meta announced the 'Llama Impact Challenge' - an initiative to inspire a diverse group of public, non-profit, and for-profit entities to use Llama 2 in addressing environmental, educational, and other significant challenges. Details about the challenge rules are expected to be released shortly.