Improve memory usage by properly cleaning up weights as quantized #16
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was
linked to
issues
Jun 13, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Before this PR, we would run out of memory quantizing the weights of Llama 3 70B on two H100 80GB GPUs.
This is because as we were quantizing the weights, we were holding references to the original Linear modules such that PyTorch wouldn't free everything. Now we explicitly clone the weights and biases to then delete all of the original Parameters, as we quantize each module. This seems to massively improve peak memory usage and essentially makes it not an issue beyond the initial unquantized checkpoint load.