]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: b143fbc)
convert : force patch_embd weights to F16 or F32 to avoid broken GGUFs (#15367)
authorSigbjørn Skjæret <redacted>
Sun, 17 Aug 2025 12:47:42 +0000 (14:47 +0200)
committerGitHub <redacted>
Sun, 17 Aug 2025 12:47:42 +0000 (14:47 +0200)
commit4d196981d4db79e0105b939eaa7ecd40385b721c
treef34e49094969899e6a2afac0e5c5be61e043758atree
parentb143fbc87af0324aa49e16cd91faf3ba8bb22231commit | diff
convert : force patch_embd weights to F16 or F32 to avoid broken GGUFs (#15367)

* force patch_embd weights to f32

* use MmprojModel base tensor_force_quant instead
convert_hf_to_gguf.py diff | blob | history
Packaging of ggml-org/llama.cpp