Skip to content

LLM inference in C/C++

C++ ggml

ggml-cpu : fix missing Q5_0 dispatch in SpaceMiT backend (#26792)

Hao-Chen233708659901c43b
diff --git a/ggml/src/ggml-cpu/spacemit/ime.cpp b/ggml/src/ggml-cpu/spacemit/ime.cpp
index 9563ea3e4b..29d683270e 100644
--- a/ggml/src/ggml-cpu/spacemit/ime.cpp
+++ b/ggml/src/ggml-cpu/spacemit/ime.cpp
@@ -195,6 +195,7 @@ template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS> class tensor_
case GGML_TYPE_Q4_K:
case GGML_TYPE_Q6_K:
case GGML_TYPE_Q8_0:
+ case GGML_TYPE_Q5_0:
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q5_K:
//case GGML_TYPE_MXFP4:
@@ -214,6 +215,7 @@ template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS> class tensor_
case GGML_TYPE_Q4_K:
case GGML_TYPE_Q6_K:
case GGML_TYPE_Q8_0:
+ case GGML_TYPE_Q5_0:
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q5_K:
//case GGML_TYPE_MXFP4: