MiniCPM-V icon indicating copy to clipboard operation
MiniCPM-V copied to clipboard

[BUG] <title> 使用git拉取的DeepSpeed仓库按照命令pip install e . 安装的deepspeed版本号为deepspeed-0.15.4+unknown,后面运行微调命令时,遇到报错,不知道是不是和这个unknown的出现有关系

Open xueyuG opened this issue 1 year ago • 2 comments

是否已有关于该错误的issue或讨论? | Is there an existing issue / discussion for this?

  • [X] 我已经搜索过已有的issues和讨论 | I have searched the existing issues / discussions

该问题是否在FAQ中有解答? | Is there an existing answer for this in FAQ?

  • [X] 我已经搜索过FAQ | I have searched FAQ

当前行为 | Current Behavior

微调MiniCPM-V-2.6时,使用git拉取的DeepSpeed仓库按照命令pip install e . 安装的deepspeed版本号为deepspeed-0.15.4+unknown,后面运行微调命令时,遇到报错,不知道是不是和这个unknown的出现有关系 另外,报的错误中有一行为: [rank2]: pydantic_core._pydantic_core.ValidationError: 1 validation error for DeepSpeedZeroConfig 我使用的是ds_config_zero3

期望行为 | Expected Behavior

No response

复现方法 | Steps To Reproduce

No response

运行环境 | Environment

- OS: Ubuntu 20.04
- Python: 3.11
- Transformers: 4.44.0
- PyTorch: 2.4.0
- CUDA (`python -c 'import torch; print(torch.version.cuda)'`): 12.0

备注 | Anything else?

No response

xueyuG avatar Nov 14 '24 11:11 xueyuG

您好,可以黄哥vllm版本进行安装试一试

LDLINGLINGLING avatar Jan 14 '25 03:01 LDLINGLINGLING

我们这边默认的deepspeed版本是0.12.3 你可以试试

qyc-98 avatar Jan 14 '25 07:01 qyc-98