Meta Plans to Launch Smaller Llama 3 AI Models Before The Big Release

Reportedly, these smaller versions of the upcoming Llama 3 will be used to generate excitement among the users. They will be launched roughly a year after the introduction of Llama 2, which was released in July last year. Citing The Independent, a Meta employee revealed that the two smaller versions will be released in the next week.

The move to launch these smaller versions has come as many big companies like Google, Mistral, Databricks, and Elon Musk’s xAI have already launched open-sourced LLMs in the past. Meta officials said, “The models will serve as a precursor to the launch of the biggest version of Llama 3, expected this summer.”As large models are extremely useful as they are powerful and there are very few chances of mistakes when handling sophisticated tasks,they become quite costly. On the other hand, smaller models are cost-effective and have a faster processing speed. Smaller models require less computational power and memory to run, making them faster to run on smaller devices like mobile phones.

These smaller models are also quite flexible in their deployment into other devices, making them easier to integrate into already existing systems and applications without causing much disruption in their systems.

Companies like Google have already launched small versions of their AI models, like the Gemini Nano, which is used for mobile applications. Apart from that, they also launched other smaller versions, including Gemini Pro, used for scaling different types of tasks, and Gemini Ultra, which is Google’s top-of-the-line model.

Talking about Llama 3, it is part of Meta’s strategy to counter OpenAI’s GPT-4, which is known for answering people’s questions based on user-uploaded pictures. Llama 3 will be equipped with the ability to process both images and texts, but the two smaller versions will not have these capabilities.