Mitigating Quantization Errors Due to Activation Spikes in Gated Linear Unit-Based Large Language Models.

Jaewoo Yang, Hayun Kim, Junyung Ji, Younghoon Kim