Skip to content

trainer删掉了epoch的逻辑 #49

@wjdmyby98k

Description

@wjdmyby98k

您好,我看了部分源码里面的trainer,发现都是改掉了epoch的逻辑,按照step去训练,那如果我想还是按照epoch去训练,应该怎么修改呢?其次是weighter模块貌似默认是关闭liger kernel的,但是一般训练都会开启liger kernel加快训练速度,然而开启以后会有一些优化,拿不到每个样本的loss,只有每个batch的loss,weighter内部会直接不生效,这块有考虑支持一下batch weighter嘛,感谢!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions