Comment Comment by applicative "At a standard Q4_K_M quantization, the model footprint takes about 459.5 GB." parent applicative 21 days ago
