Skip to content

make weight override more robust against ggml changes #760

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

make weight override more robust against ggml changes #760

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1c30154
cleanup: do not force VAE type to f32 on SDXL
wbruna Aug 7, 2025
f9a6c7f
chore: make weight override more robust against ggml changes
wbruna Aug 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2156,7 +2156,7 @@ std::vector<std::pair<std::string, ggml_type>> parse_tensor_type_rules(const std
if (type_name == "f32") {
tensor_type = GGML_TYPE_F32;
} else {
for (size_t i = 0; i < SD_TYPE_COUNT; i++) {
for (size_t i = 0; i < GGML_TYPE_COUNT; i++) {
auto trait = ggml_get_type_traits((ggml_type)i);
if (trait->to_float && trait->type_size && type_name == trait->type_name) {
tensor_type = (ggml_type)i;
Expand Down
16 changes: 8 additions & 8 deletions stable-diffusion.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -241,7 +241,9 @@ class StableDiffusionGGML {
}

LOG_INFO("Version: %s ", model_version_to_str[version]);
ggml_type wtype = (ggml_type)sd_ctx_params->wtype;
ggml_type wtype = (int)sd_ctx_params->wtype < std::min<int>(SD_TYPE_COUNT, GGML_TYPE_COUNT)
? (ggml_type)sd_ctx_params->wtype
: GGML_TYPE_COUNT;
if (wtype == GGML_TYPE_COUNT) {
model_wtype = model_loader.get_sd_wtype();
if (model_wtype == GGML_TYPE_COUNT) {
Expand Down Expand Up @@ -269,11 +271,6 @@ class StableDiffusionGGML {
model_loader.set_wtype_override(wtype);
}

if (sd_version_is_sdxl(version)) {
vae_wtype = GGML_TYPE_F32;
model_loader.set_wtype_override(GGML_TYPE_F32, "vae.");
}

LOG_INFO("Weight type: %s", model_wtype != GGML_TYPE_COUNT ? ggml_type_name(model_wtype) : "??");
LOG_INFO("Conditioner weight type: %s", conditioner_wtype != GGML_TYPE_COUNT ? ggml_type_name(conditioner_wtype) : "??");
LOG_INFO("Diffusion model weight type: %s", diffusion_model_wtype != GGML_TYPE_COUNT ? ggml_type_name(diffusion_model_wtype) : "??");
Expand Down Expand Up @@ -1216,11 +1213,14 @@ class StableDiffusionGGML {
#define NONE_STR "NONE"

const char* sd_type_name(enum sd_type_t type) {
return ggml_type_name((ggml_type)type);
if ((int) type < std::min<int>(SD_TYPE_COUNT, GGML_TYPE_COUNT)) {
return ggml_type_name((ggml_type)type);
}
return NONE_STR;
}

enum sd_type_t str_to_sd_type(const char* str) {
for (int i = 0; i < SD_TYPE_COUNT; i++) {
for (int i = 0; i < std::min<int>(SD_TYPE_COUNT, GGML_TYPE_COUNT); i++) {
auto trait = ggml_get_type_traits((ggml_type)i);
if (!strcmp(str, trait->type_name)) {
return (enum sd_type_t)i;
Expand Down
Loading