Mitigating the Impact of Outlier Channels for Language Model Quantization with Activation Regularization

Open in new window