Skip to content

gguf_convert_endian.py: add support for converting BF16 data #16594

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 15, 2025
Merged

gguf_convert_endian.py: add support for converting BF16 data #16594

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions gguf-py/gguf/scripts/gguf_convert_endian.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,7 @@ def convert_byteorder(reader: gguf.GGUFReader, args: argparse.Namespace) -> None
tensor.tensor_type not in (
gguf.GGMLQuantizationType.F32,
gguf.GGMLQuantizationType.F16,
gguf.GGMLQuantizationType.BF16,
):
raise ValueError(f"Cannot handle type {tensor.tensor_type.name} for tensor {repr(tensor.name)}")
logger.info(f"* Preparing to convert from {file_endian} to {order}")
Expand Down Expand Up @@ -148,6 +149,11 @@ def convert_byteorder(reader: gguf.GGUFReader, args: argparse.Namespace) -> None

# restore old shape in case it's ever used
tensor.data.resize(oldshape)
elif tensor.tensor_type == gguf.GGMLQuantizationType.BF16:
# Special case for BF16
# It is 2-bytes data, but by default view loads it as 1-byte data.
# Change to correct view before byteswapping.
tensor.data.view(dtype=np.uint16).byteswap(inplace=True)
else:
# Handle other tensor types
tensor.data.byteswap(inplace=True)
Expand Down