Skip to content

mag_bert:Train loss NaN #16

@Jerry-xie-22

Description

@Jerry-xie-22

您好,我按照readme指引运行代码,在运行text方法时,是可以正常训练和测试的。但是发现在运行mag_bert方法时,CTC对齐以后输出的视频和语音特征的Tensor值大部分为NaN,训练时loss一直为NaN,请问有解决办法嘛?

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