Vector Quantization With Self-Attention for Quality-Independent Representation Learning | Read Paper on Bytez