Welcome Llama 3 - Meta's new open LLM
Hugging Face Blog
Meta released Llama 3, an open-source large language model available in 8B and 70B parameter sizes, along with Llama Guard 2 for content safety classification. The models were trained on over 15 trillion tokens using 24,000 GPUs and feature a expanded tokenizer vocabulary of 128,256 tokens compared to 32K in Llama 2. The release includes integrations with Hugging Face Transformers, Google Cloud, Amazon SageMaker, and deployment options for both consumer GPUs and enterprise infrastructure.
