Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

微调时显存都占得很满 #1314

Open
fly-dragon211 opened this issue Jul 7, 2024 · 1 comment
Open

微调时显存都占得很满 #1314

fly-dragon211 opened this issue Jul 7, 2024 · 1 comment

Comments

@fly-dragon211
Copy link

您好,感谢 swift 很方便的框架,我发现微调时无论怎么设置参数,显存都占得很满,感觉随时都可能爆显存,请问这是怎么回事呢?

Sun Jul  7 23:21:53 2024  470.239.06
[0] Tesla V100-SXM2-32GB | 68°C,  84 % | 32507 / 32510 MB |
[1] Tesla V100-SXM2-32GB | 71°C, 100 % | 32381 / 32510 MB |
[2] Tesla V100-SXM2-32GB | 64°C,  61 % | 32377 / 32510 MB |
[3] Tesla V100-SXM2-32GB | 71°C, 100 % | 32269 / 32510 MB |
[4] Tesla V100-SXM2-32GB | 70°C, 100 % | 32495 / 32510 MB |
[5] Tesla V100-SXM2-32GB | 77°C, 100 % | 32257 / 32510 MB |
[6] Tesla V100-SXM2-32GB | 68°C, 100 % | 32233 / 32510 MB |
[7] Tesla V100-SXM2-32GB | 76°C, 100 % | 32175 / 32510 MB |

我运行的脚本是官方训练 qwen-vl 的脚本

@tastelikefeet
Copy link
Collaborator

pytorch在进行训练时会产生缓存,训练时显存释放周期随显卡的显存而定
我这边实际试了下,感觉大概占用22G,考虑是否是因为不支持flash-attn导致的显存下不来

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants