Comment by shad0wca7

Since the quantization requires more ram than running it, why can’t the quantized models be uploaded for use by those with 16-32gb?