Skip to content

LLM inference in C/C++

C++ ggml

fix(mtmd): handle Gemma 4 audio projector embedding size (#24091)

Andreie3ba22d6cc4d
diff --git a/tools/mtmd/clip.cpp b/tools/mtmd/clip.cpp
index 2008198150..80136ed866 100644
--- a/tools/mtmd/clip.cpp
+++ b/tools/mtmd/clip.cpp
@@ -4347,6 +4347,8 @@ int clip_n_mmproj_embd(const struct clip_ctx * ctx) {
return ctx->model.mm_input_proj_w->ne[0];
case PROJECTOR_TYPE_GEMMA4V:
case PROJECTOR_TYPE_GEMMA4UV:
+ case PROJECTOR_TYPE_GEMMA4A:
+ case PROJECTOR_TYPE_GEMMA4UA:
return ctx->model.mm_input_proj_w->ne[1];
case PROJECTOR_TYPE_IDEFICS3:
return ctx->model.mm_fc_w->ne[1];
@@ -4381,8 +4383,6 @@ int clip_n_mmproj_embd(const struct clip_ctx * ctx) {
return ctx->model.mm_fc_w->ne[1];
case PROJECTOR_TYPE_LFM2A:
return ctx->model.position_embeddings->ne[0];
- case PROJECTOR_TYPE_GEMMA4UA:
- return ctx->model.hparams.projection_dim;
case PROJECTOR_TYPE_GRANITE_SPEECH:
return ctx->model.qf_proj_linear_w->ne[1];
case PROJECTOR_TYPE_GLM4V: