http://zh.d2l.ai/chapter_attention-mechanisms/attention-scoring-functions.html
1 Like
加性注意力里 query 和 keys 的相似度该怎么理解？ 点积注意力很好理解，点积本身就是相似度。
how to understand the similarity between query and kyes in additive attention pooling？
i understand scaled-dot product attention， it likes cosine similarity。