HOPE:面向 MoE 大语言模型的高阶专家剪枝方法
Your Experts Have Partners. There's HOPE for Keeping Them Together
AI 导读
Hugging Face 团队提出 HOPE(Higher-Order Pruning of Experts),一种考虑专家两两交互的二阶剪枝方法,用于压缩 MoE 大语言模型。
Your Experts Have Partners. There's HOPE for Keeping Them Together
Hugging Face 团队提出 HOPE(Higher-Order Pruning of Experts),一种考虑专家两两交互的二阶剪枝方法,用于压缩 MoE 大语言模型。