LLama 3 — World best Open Source AI model

Amber Ivanna Trujillo
5 min readApr 21, 2024

Meta (formerly Facebook) has invested over $100,000,000 to develop Llama 3, the world’s leading open-source AI model.

  • Users can access Llama 3 for free, thanks to Meta’s commitment to open-source AI development.
  • Meta stands out among Big Tech companies for its focus on creating large language models with an open-source approach.
  • The primary motivation behind Meta’s investment in Llama 3 appears to be strategic rather than purely altruistic.
  • By providing Llama 3 at no cost, Meta aims to dominate the market and eliminate commercial competition in the AI field.
  • Meta launched Llama 3 Large Language Models (LLMs) on April 18th, 2024, in 8B and 70B parameter sizes, optimized for dialogue use cases. These models, trained on over 15 trillion tokens, utilize Grouped-Query Attention for improved scalability. With supervised fine-tuning and reinforcement learning, they align with human preferences for safety and helpfulness. Llama 3 supports a context length of 8K tokens and is intended for commercial and research use in English. Evaluated for cybersecurity safety, Llama 3 proved comparable or safer than models of similar capability. Trained on Nvidia GPUs, these models consumed approximately $16+ Million.

Llama 3: Meta’s Cutting-Edge AI Model

--

--

Amber Ivanna Trujillo

I am Executive Data Science Manager. Interested in Deep Learning, LLM, Startup, AI-Influencer, Technical stuff, Interviews and much more!!!