In the battle to win over users with artificial intelligence, another giant has broken out between Microsoft and Google. Meta wants to gain a place on the podium in this hectic race and is already showing the amazing results of CM3leon (pronounced chameleon, chameleon in English). Image generator from Mark Zuckerberg’s company also interprets them and can edit them in any way you ask. He works from them or creates them, as requested by the user. Meta’s week in the field of AI ends with the announcement that his LLaMA 2 model has become open source, so anyone can download it to his computer and give it a specific application.
From CM3leon the date of its making available to the public is still unknown. Meta has shown its impressive results and has revealed some of its technical characteristics. “CM3leon is the first multimodal model trained with a recipe adapted from text-only linguistic models”, its researchers have explained in Meta. The authors say that “it achieves state-of-the-art performance in text-to-image generation, despite being trained with five times less computation than previous methods based on transformers – a system devised by Google for AI language models -“.
CM3leon also uses one of the usual methods in AI image generation, which consists of “adding a separately trained super resolution stage to produce higher resolution images from the original model results”. The results shown by Meta are of quality, although there are no clues as to when or under what conditions it will be available to the public.
In any case, Meta is inclined to open up access to its artificial intelligence models very much. On Tuesday, he announced with Microsoft – the largest investor in OpenAI, which is bringing ChatGPT’s AI to all of its commercial products – an agreement to make Mark Zuckerberg’s company’s artificial intelligence, LLaMa 2, available as open source for anyone who wants to use it to develop applications.
The announcement of the free availability of LLaMA 2 was made during Microsoft’s Inspire event, which the Redmond company dedicates to the companies with which it partners around the world, because its AI model will be distributed through the Azure platform, but also from other important ones such as Amazon Web Services. Meta’s move aims to make a large language model available to developers for free, unlike other closed ones like OpenAI’s GPT-4, only available to its partners. Meta expects to grow in users from this open source strategy.
Some of the features that Meta has unveiled of its great AI language model far surpass the previous one, LLaMA 1. The new one has been trained with 40% more data than the first one and has twice the context extension than this one.
LLaMA 2 is now available to anyone with three different sizes that are measured according to the number of parameters used by the language model: 7,000, 13,000 and 70,000 million. This latest version is comparable to GPT-3.5, the model with which OpenAI launched ChatGPT last November, although now it also has a greatly improved model, whose features are barely known, GPT-4, which has the ability to be more in-depth in its responses and is multimodal, capable of using text and images.
Meta ensures that its AI models “have undergone internal and external security tests.” The company also provides developers with a guide on responsible practices and security assessments. The EU is preparing a law that will regulate this technology.
