Mark Zuckerberg

Mark Zuckerberg to release Meta’s new large language model called LLaMA

Mark Zuckerberg, the co-founder of Facebook, announced Meta Platforms’ impending release of LLaMA, Meta’s new large language model, to researchers (Large Language Model Meta AI). The model, created by Meta’s Fundamental AI Research team, is intended to help scientists and engineers investigate AI applications and functions such as answering questions and summarising documents.

The release of LLaMA coincides with a race among technology companies to promote advances in AI techniques and incorporate technology into their commercial products. It includes applications like Microsoft’s Bing AI, OpenAI’s ChatGPT, and Google’s Bard. Meta’s release differs from competitors’ models in that it will be available in a variety of sizes ranging from 7 billion parameters to 65 billion parameters. Furthermore, Zuckerberg stated that his company’s new LLM technology, which could eventually solve math problems and conduct scientific research, will be made available to the research community, and Meta is currently accepting access applications.

Meta is joining a growing race to dominate AI technology, which began in earnest in late 2022 with OpenAI’s ChatGPT. In terms of Meta, the launch of LLaMA also represents the company’s commitment to open science. Unlike more finely tuned models designed for specific purposes, Meta believes theirs will be versatile, with multiple use cases. According to reports, Meta’s LLaMA, short for Large Language Model Meta AI, will be available under a non-commercial license to researchers and entities affiliated with the government, civil society, and academia. Another way LLaMA differs is that it requires “far less” computing power than previous offerings and is trained in 20 languages, with a focus on those based on the Latin and Cyrillic alphabets. LLaMA should outperform GPT-3, the model on which ChatGPT is based, with its 13 billion parameters. Meta also attributed LLaMA’s success to “cleaner” data and model “architectural improvements” that improved training stability. According to reports, Meta’s LLaMA, short for Large Language Model Meta AI, will be available under a non-commercial license to researchers and entities affiliated with the government, civil society, and academia.  In May of 2022, Meta released the large language model OPT-175B, which was also aimed at researchers and served as the foundation for a new iteration of its chatbot BlenderBot.