vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Add more quantization types to unsupported

Browse source
This commit is contained in:
Aidan 2024-02-27 12:34:38 +00:00
parent cbbd1efa06
commit 58eca1db04
1 changed files with 13 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

14
ggml-sycl.cpp
View file

@ -12010,7 +12010,10 @@ inline void ggml_sycl_op_mul_mat_vec_q(
GGML_ASSERT(!(src0->type == GGML_TYPE_IQ2_XXS ||
src0->type == GGML_TYPE_IQ2_XS ||
src0->type == GGML_TYPE_IQ3_XXS ||
src0->type == GGML_TYPE_IQ1_S));
src0->type == GGML_TYPE_IQ1_S ||
src0->type == GGML_TYPE_IQ2_S ||
src0->type == GGML_TYPE_IQ3_S ||
src0->type == GGML_TYPE_IQ4_NL));
switch (src0->type) {
case GGML_TYPE_Q4_0:
@ -15051,6 +15054,15 @@ static bool ggml_backend_sycl_supports_op(ggml_backend_t backend, const ggml_ten
if (a->type == GGML_TYPE_IQ1_S) {
return false;
}
if (a->type == GGML_TYPE_IQ2_S) {
return false;
}
if (a->type == GGML_TYPE_IQ3_S) {
return false;
}
if (a->type == GGML_TYPE_IQ4_NL) {
return false;
}
if (a->type == GGML_TYPE_IQ3_XXS) {
return false;
}