Skip to content

[RWKV-v5] use register_buffer instead of frozen params#213

Open
kashif wants to merge 1 commit into
BlinkDL:mainfrom
kashif:buffer
Open

[RWKV-v5] use register_buffer instead of frozen params#213
kashif wants to merge 1 commit into
BlinkDL:mainfrom
kashif:buffer

Conversation

@kashif

@kashif kashif commented Dec 19, 2023

Copy link
Copy Markdown

also fixed a bug due to this in MishGLU

@BlinkDL

BlinkDL commented Dec 20, 2023

Copy link
Copy Markdown
Owner

these parameters are trainable
so we use nn.Parameter

@kashif

kashif commented Dec 20, 2023

Copy link
Copy Markdown
Author

ah ok then what does the with torch.no_grad(): do?

@kashif

kashif commented Dec 29, 2023

Copy link
Copy Markdown
Author

@BlinkDL happy to close this PR or should i remove the torch.no_grad()?

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

Successfully merging this pull request may close these issues.

2 participants