Skip to content

Implement parallelization over heads #2

@certik

Description

@certik

Currently the attention over heads runs in serial:

! Perform attention over each head

We should try to parallelize it and see if we get any speedups.

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