Running Llama2 on a 50-Tflop AMD GPU Inference Server for $400

yehowshuaimmanuel.com