Allow safetensors offload #873
Merged
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
|
The documentation is not available anymore as the PR was closed or merged. |
muellerzr
reviewed
Nov 18, 2022
src/accelerate/utils/offload.py
Outdated
Comment on lines
163
to
164
| if weight_info.get("safetensors_file") is not None: | ||
| from safetensors import safe_open |
There was a problem hiding this comment.
We need a guard here for is_safetensors_available and raise an error if not :) (aka make one :) )
There was a problem hiding this comment.
Indeed, will clean that up on Monday!
muellerzr
approved these changes
Nov 28, 2022
4 tasks
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Outdated suggestions cannot be applied.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.
This PR reworks a bit the internals of the offload hooks to allow for the case where the checkpoint is saved with
safetensors. In this case, there is no need to have a separate save of the weight as a Numpy memory-mapped array, as we can directly access the weight usingsafe_open.Goes along with huggingface/transformers#20321