Comment by MacsHeadroom

I'm running LLaMA-65B on a single A100 80GB with 8bit quantization. $1.5/hr on vast.ai