Tag - Transformer

从 Input Embeddings 到 Context Vectors

Bias Vector

Understanding the role of bias vectors in multi-head attention mechanisms.

因果自注意力与多头注意力

因果自注意力和高效多头注意力模块的PyTorch实现

Single Head Self-Attention

Single Head Self-Attention的完整实现

一个简单的Self-Attention机制的实现

一个简单的自注意力机制实现

trainable weight self Attention

simple self attention -> trainable weight self Attention

Greedy search decoding

一个Greedy search decoding的例子