Add Llama-3-8B-Instruct-GPTQ-4-Bit model without embedded repository

initial commit