Llama 3 70B framework is designed to enable efficient training of large-scale language models on AMD GPUs
