]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 3e5036f)
Fix stale tensor-split params for draft models (#24814)
authorAl G <redacted>
Sun, 5 Jul 2026 18:39:36 +0000 (19:39 +0100)
committerGitHub <redacted>
Sun, 5 Jul 2026 18:39:36 +0000 (20:39 +0200)
commit2da668617612d2df773f966e3b0ee22dc2beef7b
treec2361ac89d2a3aa2ba94e7b8ae15ab13ef0e41d9tree
parent3e5036fbfb1613dc19034844c1989aa645183164commit | diff
Fix stale tensor-split params for draft models (#24814)

* meta: fix tensor split metadata for GQA attention

* Tidied the code a bit to match existing style

* Revert "Tidied the code a bit to match existing style"

This reverts commit b90c6c6300091fe09e2350a3d4edcfcf15db8d2e.

* Reverted the ggml-backend-meta asset hack.
src/llama-model.cpp diff | blob | history
Packaging of ggml-org/llama.cpp