Hyun-Min Chang

Papers that credit Hyun-Min Chang in their source bylines. Matching names may refer to different researchers.

1 results

1. PaperMachine Learning

   MACaQ keeps four-bit quantization within each layer

   MACaQ keeps four-bit neural-network quantization layer-local, with better text-prediction scores and faster quantization than MR-GPTQ in one Llama-3-8B setup.

   Added to Chiosa Nov 12, 2026