Fetching the paper…

Mitigating Quantization Errors Due to Activation Spikes in GLU-Based LLMs · Around