torchao: add int8; quanto: add NF4; torch compile fixes + ability to compile optim #986
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
added 8 commits
September 23, 2024 21:06
|
@sayakpaul i can't get this to work on NVIDIA devices, but it does work on Apple MPS. have you had any luck training on torchao? for cuda it complains it cannot cast (tensor, tensor) to tensor. |
|
currently the main benefit of torchao i can see is that we get torch compile support, but torch compile doesn't even work on mps.. on MPS, torchao is 3 seconds per step slower than quanto, which is roughly the same speed as bf16, tested on Flux Dev. so if there's no nvidia support for training and it works more slowly on MPS, not sure there is a point to merge this PR |
This comment was marked as outdated.
This comment was marked as outdated.
|
i change my mind, i got it to work with int8 on nvidia after messing around with the internals on the GPUMODE discord. |
bghira
changed the title
bghira
commented
Sep 28, 2024
…int8 when dynamo is enabled
bghira
changed the title
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Outdated suggestions cannot be applied.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.
No description provided.