ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation

ViTPose：用于人体姿态估计的简单视觉Transformer基线.

PolyLoss: A Polynomial Expansion Perspective of Classification Loss Functions

PolyLoss：一种分类损失函数的多项式展开视角.

RoFormer: Enhanced Transformer with Rotary Position Embedding

RoFormer：使用旋转位置编码增强Transformer.

Your Transformer May Not be as Powerful as You Expect

使用通用相对位置编码改进Transformer的通用近似性.

Encoding word order in complex embeddings

在复数域空间中构造词嵌入.

Self-Attention with Relative Position Representations

自注意力机制中的相对位置编码.