Skip to content

Conversation

@dimakarp1996
Copy link
Contributor

Added 8bit optimizers. These optimizers, as my experiments have shown, yield ~9-11% memory economy on training. The code is absolutely backward-compatible.

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.

3 participants