ChatGLM-6B icon indicating copy to clipboard operation
ChatGLM-6B copied to clipboard

[BUG/Help] <加载模型出错>

Open JJDawn opened this issue 1 year ago • 1 comments

Is there an existing issue for this?

  • [X] I have searched the existing issues

Current Behavior

image

在执行“代码调用”相关代码的时候,每次加载到Loading checkpoint shards: 38%都会Killed。

image

包括手动下载参数,再加载到模型依旧会出现该问题: image image image

Expected Behavior

想问一下,不论是Transformer加载还是本地加载,都需要执行Loading checkpoint shards吗。如果不是,可能是什么原因造成的呢?内存吗?

Steps To Reproduce

1.OS: WSL-Ubuntu 2.python: 3.9

Environment

- OS:WSL-Ubuntu
- Python:3.9
- Transformers:4.27.1
- PyTorch:2.0
- CUDA Support (`python -c "import torch; print(torch.cuda.is_available())"`) :

Anything else?

No response

JJDawn avatar May 04 '23 11:05 JJDawn

可能是内存不够,可以运行的过程中执行free看下,是不是kill掉的时候内存超了。

tykuyh avatar May 04 '23 11:05 tykuyh