模型剪枝经典论文精度:Are Sixteen Heads Really Better than One?
一、论文基本信息
论文题目:Are Sixteen Heads Really Better than One?
作者:Paul Michel、Omer Levy、Graham Neubig
发表会议:NeurIPS 2019
研究方向:Transformer 模型压缩、Attention Head Pruning、多头注意力结构分析 …
2026/7/18 4:38:08