Skip to content

华为昇腾使用accelerate调度器微调qwen3-next-80B,但是在配置文件中deepspeed_moe_layer_cls_names不知道应该写什么? #10249

@gongyinfeng0206-design

Description

@gongyinfeng0206-design

Reminder

  • I have read the above rules and searched the existing issues.

System Info

华为昇腾910B4,32G*8,单机8卡
accelerate 1.11.0
deepspeed 0.16.9
llamafactory 0.9.5

Reproduction

Image

Image

Image

Others

查找源代码是从环境变量中加载的,但是输出该环境变量是空的?
不知道在微调时需要怎么写这个参数?注释掉是不是就没有显存优化?

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingnpuThis problem is related to NPU devicespendingThis problem is yet to be addressed

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions