]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: d35a1e8)
gguf-py: byteswapping improvements (#12851)
authorAleksei Nikiforov <redacted>
Thu, 28 Aug 2025 08:56:41 +0000 (10:56 +0200)
committerGitHub <redacted>
Thu, 28 Aug 2025 08:56:41 +0000 (16:56 +0800)
commit64387f6e95434b393ac3df285864692b7fd9c4d2
tree08c74f74263c178561000906a06a1bcc55599717tree
parentd35a1e8c41f747548775225973a99507896a8c61commit | diff
gguf-py: byteswapping improvements (#12851)

* gguf-py: implement byteswapping for Q4_0

This is needed to byteswap Mistral model.

Also restore original shapes after byteswapping tensors.
It is not needed at the moment, but do it in case
they'd be used in future.

* Rework byteswapping code in gguf-py

Move out details from byteswapping tensor blocks code
gguf-py/gguf/scripts/gguf_convert_endian.py diff | blob | history
Packaging of ggml-org/llama.cpp