Skip to content

qwenvl 精度阈值问题 #95

@CokeDong

Description

@CokeDong

厂商反馈:qwen2.5vl的loss base,readme是以GBS64 TP1PP1DP8, gradient accumulation steps是8
在GBS64 TP1PP4DP2, gradient accumulation steps是32,会和readme的base有些差异(不在5%以内了)

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