• UGI USER
  • Apr 23, 2024
  • 19 Likes
  • 0 Comments

“META UNVEILS LLAMA 3, ITS MOST ADVANCED LLM MODEL.”

Why in the news?

“Meta introduces Llama 3, its latest LLM, boasting superior performance and diverse applications in AI chatbots and language tasks.”

Key Features of Llama 3 :

  • Llama 3, successor to Llama 2, comes in two sizes: 8 billion and 70 billion parameters, enhancing model complexity.
  • Supports context lengths of 8,000 tokens, enabling more interactions and complex input handling compared to predecessors.
  • Boasts major improvements in pre-training and post-training processes, enhancing capabilities like reasoning and code generation.
  • Meta claims Llama 3 outperforms other open-source LLMs in benchmark evaluations, offering superior performance in various tasks.
  • While specific use cases are not officially stated, Llama 3 can generate diverse text forms like poems, code, and scripts, summarise factual topics, and translate languages.
source:wordpress
About Llama 3:

●    Based on Llama 2 architecture.

●    Released in 8B and 70B parameter sizes.

●    Includes base model and instruction-tuned version for specific tasks.

Characteristics:

●    Significant performance progress, enhanced AI capabilities.

●    Used for generating various text forms, summarising topics, and language translation.

●    Powers AI chatbots for conversational interactions.

●    Embraces open-source ethos, offers access during development.

Integration and Safety:

●    Integrated into Meta AI for platforms like Facebook, Instagram, and others.

●    Accessible via Hugging Face ecosystem, Perplexity Labs, Fireworks AI, Azure ML, and Vertex AI.

●   Meta emphasises responsible development, offers safety tools like Llama Guard 2 and Code Shield.

Source:Indian Express