diff options
Diffstat (limited to 'ggml-cuda.cu')
| -rw-r--r-- | ggml-cuda.cu | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/ggml-cuda.cu b/ggml-cuda.cu index 4f50c9f9..48232b6e 100644 --- a/ggml-cuda.cu +++ b/ggml-cuda.cu @@ -615,6 +615,7 @@ static int64_t get_row_rounding(ggml_type type, const std::array<float, GGML_CUD case GGML_TYPE_IQ2_S: case GGML_TYPE_IQ3_XXS: case GGML_TYPE_IQ1_S: + case GGML_TYPE_IQ1_M: case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ3_S: @@ -643,6 +644,7 @@ static int64_t get_row_rounding(ggml_type type, const std::array<float, GGML_CUD case GGML_TYPE_IQ2_S: case GGML_TYPE_IQ3_XXS: case GGML_TYPE_IQ1_S: + case GGML_TYPE_IQ1_M: case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ3_S: @@ -2560,7 +2562,7 @@ GGML_CALL static bool ggml_backend_cuda_supports_op(ggml_backend_t backend, cons ggml_type a_type = a->type; if (a_type == GGML_TYPE_IQ2_XXS || a_type == GGML_TYPE_IQ2_XS || a_type == GGML_TYPE_IQ3_XXS || a_type == GGML_TYPE_IQ1_S || a_type == GGML_TYPE_IQ4_NL || a_type == GGML_TYPE_IQ3_S || - a_type == GGML_TYPE_IQ2_S || a_type == GGML_TYPE_IQ4_XS) { + a_type == GGML_TYPE_IQ1_M || a_type == GGML_TYPE_IQ2_S || a_type == GGML_TYPE_IQ4_XS) { if (b->ne[1] == 1 && ggml_nrows(b) > 1) { return false; } |