1 paper
Angelos Korakitis, Vasileios Nakos
Quantization has emerged as one of the primary techniques for reducing the memory footprint of large language models during inference. A prominent example is Quantized Johnson−Li…