fix(utils): jax_to_pytorch NumPy bridge cannot convert bfloat16 arrays#269
Open
opooladz wants to merge 1 commit into
Open
fix(utils): jax_to_pytorch NumPy bridge cannot convert bfloat16 arrays#269opooladz wants to merge 1 commit into
opooladz wants to merge 1 commit into
Conversation
torch.from_numpy raises TypeError for ml_dtypes.bfloat16 inputs (NumPy has no native bfloat16), so the default EASY_SAFE_TRANSFER path failed for any bf16 model export (module.to_torch / save_pretrained(to_torch= True)) with: "can't convert np.ndarray of type ml_dtypes.bfloat16". Reinterpret the array bits as uint16 and view the resulting tensor back as torch.bfloat16 -- bit-exact (verified by raw-bit comparison in the added tests), no upcast, no additional copy. Other dtypes take the existing path unchanged.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
torch.from_numpy raises TypeError for ml_dtypes.bfloat16 inputs (NumPy has no native bfloat16), so the default EASY_SAFE_TRANSFER path failed for any bf16 model export (module.to_torch / save_pretrained(to_torch= True)) with: "can't convert np.ndarray of type ml_dtypes.bfloat16".
Reinterpret the array bits as uint16 and view the resulting tensor back as torch.bfloat16 -- bit-exact (verified by raw-bit comparison in the added tests), no upcast, no additional copy. Other dtypes take the existing path unchanged.