Description
Calculates the attention score based on the traditional flash attention mechanism that manages kv cache based on the layer ID.
Parent topic: SelfAttentionOperation
我们使用cookie来确保您的高速浏览体验。继续浏览本站，即表示您同意我们使用cookie。查看详情
Calculates the attention score based on the traditional flash attention mechanism that manages kv cache based on the layer ID.