[core] Fix safetensors serialization for shared tensors
#1101
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. |
younesbelkada
commented
Nov 9, 2023
Comment on lines
+212
to
+216
| if isinstance(tensor, torch.Tensor): | ||
| ptrs[id_tensor_storage(tensor)].append(name) | ||
| else: | ||
| # In the non-tensor case, fall back to the pointer of the object itself | ||
| ptrs[id(tensor)].append(name) |
There was a problem hiding this comment.
Not sure if this is needed, I added that in the transformers code to handle bnb state dict case and should be safe to keep it but happy to remove it as well
There was a problem hiding this comment.
Probably it's better to follow the same steps as transformers.
BenjaminBossan
approved these changes
Nov 9, 2023
| @@ -199,6 +203,25 @@ def save_pretrained( | |||
| os.makedirs(output_dir, exist_ok=True) | |||
|
|
|||
| if safe_serialization: | |||
| # Safetensors does not allow tensor aliasing. | |||
There was a problem hiding this comment.
This block is basically copied from https://github.com/huggingface/transformers/blob/main/src/transformers/modeling_utils.py#L2111-L2134, right? Let's add a comment.
Comment on lines
+212
to
+216
| if isinstance(tensor, torch.Tensor): | ||
| ptrs[id_tensor_storage(tensor)].append(name) | ||
| else: | ||
| # In the non-tensor case, fall back to the pointer of the object itself | ||
| ptrs[id(tensor)].append(name) |
There was a problem hiding this comment.
Probably it's better to follow the same steps as transformers.
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Outdated suggestions cannot be applied.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.
What does this PR do?
Fixes #1079
Safetensors do not support tensor aliasing, before saving the model state dict, we need to loop over the state dict and check for aliased tensors and
clone()them to avoid any errorcc @BenjaminBossan @pacman100