Fetching the paper…

RAPQ: Rescuing Accuracy for Power-of-Two Low-bit Post-training Quantization · Around