Fix convert_hf_to_gguf.py script on s390x#17431
Merged
CISC merged 3 commits intoggml-org:masterfrom Nov 25, 2025
Merged
Conversation
Assume converted model data is originally little-endian. Byteswap data on s390x after reading it to put values in correct presentation for any transformation needed, like calculating weight tensors. Then byteswap data to little-endian before passing it to GGUFWriter while GGUFWriter will byteswap data back to big endian if big endian output is requested. byteswap(inplace=True) calls don't work with lazy tensor and array wrappers. Use byteswap with copying data to workaround this behaviour.
…-endian With this change if no byteswapping is actually needed, 2 excessive byteswaps can be omitted on s390x
AlekseiNikiforovIBM
force-pushed
the
s390x_hf_convert
branch
from
de7b5ec to
ed94707
Compare
CISC
approved these changes
Nov 21, 2025
compilade
approved these changes
Nov 21, 2025
Comment on lines
+10049
to
+10053
| torch.uint64: np.uint64, | ||
| torch.int32: np.int32, | ||
| torch.uint32: np.uint32, | ||
| torch.int16: np.int16, | ||
| torch.uint16: np.uint16, |
Collaborator
There was a problem hiding this comment.
Might be relevant to uncomment the unsigned int types in _dtype_str_map as well (U16, U32, U64) if those are expected to exist.
They seem to be available since PyTorch 2.3.0, while the requirements.txt has version 2.6.0, so it should be fine.
Contributor
Author
There was a problem hiding this comment.
I've mentioned those types for numpy in case they're ever encountered. It would be fine for me to get _dtype_str_map updated, but maybe it could be done separately?
Contributor
Author
I don't know because I didn't test it before. |
Contributor
Author
|
Is this change ok to merge with latest commit? If yes, how do I merge it? |
Member
Yes, LGTM, I'll merge. |
Anico2
added a commit
to Anico2/llama.cpp
that referenced
this pull request
Jan 15, 2026
* Fix convert_hf_to_gguf.py script on s390x Assume converted model data is originally little-endian. Byteswap data on s390x after reading it to put values in correct presentation for any transformation needed, like calculating weight tensors. Then byteswap data to little-endian before passing it to GGUFWriter while GGUFWriter will byteswap data back to big endian if big endian output is requested. byteswap(inplace=True) calls don't work with lazy tensor and array wrappers. Use byteswap with copying data to workaround this behaviour. * Make GGUFWriter accept tensors in native endianness instead of little-endian With this change if no byteswapping is actually needed, 2 excessive byteswaps can be omitted on s390x * Fix byteswapping in convert_hf_to_gguf.py for remote models
blime4
referenced
this pull request
in blime4/llama.cpp
Feb 5, 2026
* Fix convert_hf_to_gguf.py script on s390x Assume converted model data is originally little-endian. Byteswap data on s390x after reading it to put values in correct presentation for any transformation needed, like calculating weight tensors. Then byteswap data to little-endian before passing it to GGUFWriter while GGUFWriter will byteswap data back to big endian if big endian output is requested. byteswap(inplace=True) calls don't work with lazy tensor and array wrappers. Use byteswap with copying data to workaround this behaviour. * Make GGUFWriter accept tensors in native endianness instead of little-endian With this change if no byteswapping is actually needed, 2 excessive byteswaps can be omitted on s390x * Fix byteswapping in convert_hf_to_gguf.py for remote models
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
3 participants
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.
Assume converted model data is originally little-endian.
Byteswap data on s390x after reading it to put values in correct presentation
for any transformation needed, like calculating weight tensors.
Then byteswap data to little-endian before passing it to GGUFWriter while
GGUFWriter will byteswap data back to big endian if big endian output is requested.
byteswap(inplace=True) calls don't work with lazy tensor and array wrappers.
Use byteswap with copying data to workaround this behaviour.
Make GGUFWriter accept tensors in native endianness instead of little-endian.
With this change if no byteswapping is actually needed, 2 excessive byteswaps can be omitted on s390x.