Does anyone know if vlms suffer more from quantization? One thing I've noticed is practically every improvement in llms is already half there in quantization.
Does anyone know if vlms suffer more from quantization? One thing I've noticed is practically every improvement in llms is already half there in quantization.