Skip to content

请问为什么MOE层值添加在decoder的最后一层呢,因为每一层的预测都需要经过一对一以及一对多的匹配训练,如果只在最后一层采用MOE,其他层采用普通的FFN,这样是否影响训练效果呢 #21

@yangrongkun

Description

@yangrongkun
No description provided.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions