ggml : fix I8MM Q4_1 scaling factor conversion (#10562)

ggml-ci
This commit is contained in:
Georgi Gerganov authored and GitHub committed 2024-11-29 16:25:39 +02:00
1 parent 4b3242bbea
commit f0678c5ff4
3 files changed
+38 -25

No files matched your search

+2
View File
@@ -3334,7 +3334,9 @@ static const ggml_type all_types[] = {
static const ggml_type base_types[] = {
GGML_TYPE_F32, GGML_TYPE_F16,
GGML_TYPE_Q8_0, // for I8MM tests
GGML_TYPE_Q4_0,
GGML_TYPE_Q4_1, // for I8MM tests
GGML_TYPE_Q4_K,
GGML_TYPE_IQ2_XXS
};