]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 39cffdf)
llama : fix MiniCPM inference after Granite Four changes (#14850)
authoryummy <redacted>
Thu, 24 Jul 2025 09:50:51 +0000 (17:50 +0800)
committerGitHub <redacted>
Thu, 24 Jul 2025 09:50:51 +0000 (11:50 +0200)
commit86f5623d904cfd392fdeb14a143097b4074660f6
treeb7bbd2188168709eb77a18ec403b537ca5da52d8tree
parent39cffdf18855e0d2beba62572542251d87421e73commit | diff
llama : fix MiniCPM inference after Granite Four changes (#14850)

MiniCPM models use the llm_build_granite constructor which was changed
in the Granite Four PR to use hparams.rope_finetuned instead of a
use_rope parameter. MiniCPM models need rope enabled by default.

Fixes inference from gibberish to correct responses.
src/llama-model.cpp diff | blob | history
Packaging of ggml-org/llama.cpp