Retraining-Free Model Quantization via One-Shot Weight-Coupling Learning | Read Paper on Bytez