Add empty_like for NF4Tensor to support offloading #881
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/881
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 04a5b92 with merge base 8236a87 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
drisspg
reviewed
Sep 13, 2024
janeyx99
changed the title
jerryzh168
reviewed
Sep 16, 2024
test/dtypes/test_nf4.py
Outdated
| @parametrize("input_size", [(512 * 512,), (512, 512)]) | ||
| def test_empty_like(self, input_size: Union[Tuple[int], int]): | ||
| nf4_tensor = to_nf4(torch.rand(input_size)) | ||
| new_tensor = torch.empty_like(nf4_tensor, device=torch.device("cpu")) |
There was a problem hiding this comment.
nit: to make sure device arg works, can you make sure the nf4_tensor is on gpu?
13 tasks
jainapurva
pushed a commit
that referenced
this pull request
Sep 22, 2024
Driss, we can confirm semantics when you're back but I'm fairly confident this is okay
jainapurva
pushed a commit
that referenced
this pull request
Sep 23, 2024
Driss, we can confirm semantics when you're back but I'm fairly confident this is okay
yanbing-j
pushed a commit
to yanbing-j/ao
that referenced
this pull request
Dec 9, 2024
* Rename TorchChat to Torchchat * Adding Issue templates
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Outdated suggestions cannot be applied.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.
Ran into issues when trying to offload activations when the activation was an NF4Tensor. (This was found when testing different configs for pytorch/torchtune#1443)
This PR adds
empty_likeand a test case :D The way we implementempty_likeis we traverse through the NF4Tensor's inner tensors and callempty_likeon all of them (quantized_data, quantization_factor, quantization_scalers). This may be semantically sketchy, but since we are creating garbaggio for content anyway that is expected to be overridden later, as long as we preserve the size, this seems fine.