Memory efficient 8bit backward #23
Closed
Conversation
This file contains hidden or 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
Comment on lines
+364
to
+369
| assert state.CBt is None, "CBt should not be stored in state" | ||
| CB = state.CB.half() | ||
| SCB = state.SCB.unsqueeze(1).half() | ||
| SCBt = state.SCBt.unsqueeze(1).half() | ||
| Bt = (CB * SCB).t().contiguous() | ||
| CBt = (Bt / SCBt).t().to(torch.int8) |
Contributor
There was a problem hiding this comment.
[opinion] this block of code looks like the most memory-intensive; perhaps we can optimize some of it with in-place divisions / transpositions (please ping me if you're interested)
Update bnb
Collaborator
|
This version of backward had too high of an error to be useful. As such, we will not merge it into main. See #33 for the new memory efficient tbackwards that uses fp16 computation and which seems to work fine. |
Titus-von-Koeller
pushed a commit
that referenced
this pull request
May 24, 2024
Support extract_outliers, quantize_4bit and dequantize_4bit with Device Abstraction PR.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.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.
No description provided.