We propose a novel post-training quantization method for large language models with learnable parameters, novel loss function and Test-time adaptation scheme. Post-training quantization (PTQ) for ...