GranQ: Efficient Channel-wise Quantization via Vectorized Pre-Scaling for Zero-Shot QAT

Devs

GranQ: Efficient Channel-wise Quantization via Vectorized Pre-Scaling for Zero-Shot QAT | Read Paper on Bytez