Skip to content

基于最新的LLaMA-Factory训练Qwen2.5-vl, 训练变得非常慢 #7030

Open
@leon-cas

Description

Reminder

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

System Info

llama-factory: 0.9.2.dev0
transformers: 4.49.0
python: 3.9

使用最新的LLaMA-Factory训练Qwen2-vl(2b), 采用full方式全量微调,训练速度基本无问题; 但切换成训练Qwen2.5-vl(3b)后,不管是采用full还是lora方式全量微调,训练变得异常的慢,基本比Qwen2-vl(2b)慢5-6x。

Reproduction

Put your message here.

Others

No response

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingpendingThis problem is yet to be addressed

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions