1 paper · 1 filter
Kazuki Egashira, Robin Staab, Mark Vero +2
With the increasing size of frontier LLMs, post-training quantization has become the standard for memory-efficient deployment. Recent work has shown that basic rounding-based quant…