How much vram and ram is required to run gptq_model-4bit--1g?
With Exllama I'm able to just barely load the model into 36GB of vram, ram usage is under 8GB.
Β· Sign up or log in to comment