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

mom2_dtype: "float32" #436

Open
dongdongzhaoUP opened this issue Nov 29, 2024 · 4 comments
Open

mom2_dtype: "float32" #436

dongdongzhaoUP opened this issue Nov 29, 2024 · 4 comments
Labels
question Further information is requested

Comments

@dongdongzhaoUP
Copy link

Thanks for your work. I am wondering if i can change mom2_dtype: "float32" to mom2_dtype: "float16" to save GPU memory(tried but get error).

@dongdongzhaoUP
Copy link
Author

For ROME

@pengzju
Copy link
Collaborator

pengzju commented Nov 30, 2024

应该不行

@pengzju
Copy link
Collaborator

pengzju commented Nov 30, 2024

mom2 is the precision of the inverse Key matrix. Reducing its precision would not lead to significant savings in GPU memory, and the decrease in precision might impact performance. If the GPU resources are insufficient, you can enable model_parallel.

@zxlzr zxlzr added the question Further information is requested label Nov 30, 2024
@zxlzr
Copy link
Contributor

zxlzr commented Nov 30, 2024

hi, do you have any further questions?

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

No branches or pull requests

3 participants