Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Watch: Meta's new Llama 3 models give open-source AI a boost

Apr 19, 2024 - techcrunch.com
Meta, the parent company of Facebook, has introduced two new open-source AI models as part of its Llama line of artificial intelligence tools. The Llama 3 8B and Llama 3 70B models have eight billion and seventy billion parameters respectively, with the larger number of parameters indicating a more powerful model. These models, trained on 24,000 GPU clusters, have performed well in benchmarks, outperforming some existing models in the market.

Unlike its competitors, Meta adopts an open-source approach to AI, while others, including OpenAI, prefer a more closed-source approach. This has sparked a debate in the AI world about the best approach for speed of development and safety. Despite concerns about the rapid development of AI technology and its potential risks, Meta continues to innovate, challenging its peers and rivals to match their latest offerings.

Key takeaways:

  • Meta has introduced two new AI models, Llama 3 8B and Llama 3 70B, as part of its Llama line of artificial intelligence tools. These models are open-source and have eight billion and seventy billion parameters respectively.
  • The new models were trained on 24,000 GPU clusters and have performed well in benchmarks, even outperforming some existing models in the market.
  • While Meta adopts an open-source approach to AI, its competitors, including OpenAI, often prefer a more closed-source approach. This has sparked a debate in the AI community about which approach is better for development speed and safety.
  • There are concerns that AI technology is developing too quickly and could pose a threat to democracies and other aspects of society. Despite this, Meta continues to innovate and challenge its competitors in the AI field.
View Full Article

Comments (0)

Be the first to comment!