Skip to content

Question about attention computation #13

@1190303125

Description

@1190303125

Hello, I read your paper and was impressed by the solid work. At the same time, I have a question: under the GQA settings, why are the attention scores merged before the softmax, rather than after? Before the softmax, due to the influence of the Query, the mean and variance of the attention scores might vary significantly across the different heads.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions