Comment Comment by kossTKR Does this mean that GPT-4/65b level performance is closer to running on a say a m1/m2 with only 24+ gigabytes of ram? parent kossTKR 3 years ago
