Meta Brings An Open AI Model That Supports Both Visual And Textual Content Meta’s chief executive officer (CEO) Mark Zuckerberg is in favour of open source. He is also doing some work on it. Continuing this aspect, Meta has released an open-source AI model called Llama 3.2 at the ‘Meta Connect 2024’ held on Wednesday.
It is also the first open-source AI model with visual ability. That is, this model can work on both photos and letters. This feature was not available in the open AI model provided by Meta earlier.
With the new model Lama 3.2, developers will now be able to create more modern applications in areas such as augmented reality, visual search engines, and document analysis. According to Meta, it will be easier for developers to use this new model.
Due to this, developers do not have to work hard to use the new multimodality. It is said that they can communicate with the Lama by providing a photo.
OpenAI, Google and other companies working on AI have already released multimodality models. But since photos are also supported, this AI of Meta is expected to become more impressive.
This version of Meta 3.2 includes two vision models. Which has 11 billion and 90 billion parameters. There are also two text-only models, with 1 billion and 3 billion parameters. Smaller models are designed to work with Qualcomm, MediaTek and other Arm hardware. Because of this, it is said that it can also be used on mobile phones.
