Introduction

1. Post-Training Quantization for Vision Transformer
   paper

2. Tips:

a. Similarity-Aware Quantization for Linear Operation. we adopt Pearson correlation coefficient as the measurement for the similarity

b. Ranking-Aware Quantization for Self-Attention.

c. Bias Correction

1. Similiar

a. I-BERT: Integer-only BERT Quantization: Polynomial Approximation of Non-linear Functions

b.