upstream-archive/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-07-29 14:20:57 -05:00
Code Issues Packages Projects Releases Wiki Activity

finished applying pmid lora; to be tested

Browse Source
This commit is contained in:
bssrdf
2024-02-12 21:48:57 -05:00
parent 753231a7b4
commit 44052e1478
1 changed files with 7 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
pmid.hpp
View File

@@ -547,20 +547,10 @@ struct PhotoMakerIDEncoder : public GGMLModule {
struct PhotoMakerLoraModel : public GGMLModule {
float multiplier = 1.0f;
std::map<std::string, struct ggml_tensor*> lora_tensors;
std::string file_path;
int in_channels = 4;
std::string file_path;
ModelLoader model_loader;
bool load_failed = false;
int model_channels = 320; // only for SDXL
int num_heads = -1; // only for SDXL
int num_head_channels = 64; // only for SDXL
int context_dim = 2048; // only for SDXL
PhotoMakerLoraModel(const std::string file_path = "")
: file_path(file_path) {
name = "photomaker lora";
@@ -646,8 +636,6 @@ struct PhotoMakerLoraModel : public GGMLModule {
replace_all_chars(k_tensor, '.', '_');
std::string lora_up_name = "lora." + k_tensor + ".lora_up.weight";
std::string lora_down_name = "lora." + k_tensor + ".lora_down.weight";
std::string alpha_name = "lora." + k_tensor + ".alpha";
std::string scale_name = "lora." + k_tensor + ".scale";
ggml_tensor* lora_up = NULL;
ggml_tensor* lora_down = NULL;
@@ -666,33 +654,13 @@ struct PhotoMakerLoraModel : public GGMLModule {
applied_lora_tensors.insert(lora_up_name);
applied_lora_tensors.insert(lora_down_name);
applied_lora_tensors.insert(alpha_name);
applied_lora_tensors.insert(scale_name);
// calc_cale
int64_t dim = lora_down->ne[ggml_n_dims(lora_down) - 1];
float scale_value = 1.0f;
if (lora_tensors.find(scale_name) != lora_tensors.end()) {
scale_value = ggml_backend_tensor_get_f32(lora_tensors[scale_name]);
} else if (lora_tensors.find(alpha_name) != lora_tensors.end()) {
float alpha = ggml_backend_tensor_get_f32(lora_tensors[alpha_name]);
scale_value = alpha / dim;
}
scale_value *= multiplier;
// flat lora tensors to multiply it
int64_t lora_up_rows = lora_up->ne[ggml_n_dims(lora_up) - 1];
lora_up = ggml_reshape_2d(ctx0, lora_up, ggml_nelements(lora_up) / lora_up_rows, lora_up_rows);
int64_t lora_down_rows = lora_down->ne[ggml_n_dims(lora_down) - 1];
lora_down = ggml_reshape_2d(ctx0, lora_down, ggml_nelements(lora_down) / lora_down_rows, lora_down_rows);
// ggml_mul_mat requires tensor b transposed
lora_down = ggml_cont(ctx0, ggml_transpose(ctx0, lora_down));
struct ggml_tensor* updown = ggml_mul_mat(ctx0, lora_up, lora_down);
updown = ggml_cont(ctx0, ggml_transpose(ctx0, updown));
updown = ggml_reshape(ctx0, updown, weight);
lora_up = ggml_cont(ctx0, ggml_transpose(ctx0, lora_up));
struct ggml_tensor* updown = ggml_mul_mat(ctx0, lora_down, lora_up);
updown = ggml_cont(ctx0, updown);
GGML_ASSERT(ggml_nelements(updown) == ggml_nelements(weight));
updown = ggml_scale_inplace(ctx0, updown, scale_value);
updown = ggml_scale_inplace(ctx0, updown, multiplier);
ggml_tensor* final_weight;
// if (weight->type != GGML_TYPE_F32 && weight->type != GGML_TYPE_F16) {
// final_weight = ggml_new_tensor(ctx0, GGML_TYPE_F32, weight->n_dims, weight->ne);