3 ms·At 24gb, I wonder how small the quantized models will go.by generalizations 2y agoAt 24gb, I wonder how small the quantized models will go.