upstream-archive/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-07-23 11:20:53 -05:00
Code Issues Packages Projects Releases Wiki Activity

feat: add flux support (#356)

Browse Source 
* add flux support

* avoid build failures in non-CUDA environments

* fix schnell support

* add k quants support

* add support for applying lora to quantized tensors

* add inplace conversion support for f8_e4m3 (#359)

in the same way it is done for bf16
like how bf16 converts losslessly to fp32,
f8_e4m3 converts losslessly to fp16

* add xlabs flux comfy converted lora support

* update docs

---------

Co-authored-by: Erik Scholz <Green-Sky@users.noreply.github.com>
This commit is contained in:
leejet
2024-08-24 14:29:52 +08:00
committed by GitHub
parent 697d000f49
commit 64d231f384
25 changed files with 1886 additions and 172 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
model.h
View File

@@ -18,11 +18,13 @@
#define SD_MAX_DIMS 5
enum SDVersion {
VERSION_1_x,
VERSION_2_x,
VERSION_XL,
VERSION_SD1,
VERSION_SD2,
VERSION_SDXL,
VERSION_SVD,
VERSION_3_2B,
VERSION_SD3_2B,
VERSION_FLUX_DEV,
VERSION_FLUX_SCHNELL,
VERSION_COUNT,
};
@@ -30,6 +32,7 @@ struct TensorStorage {
std::string name;
ggml_type type = GGML_TYPE_F32;
bool is_bf16 = false;
bool is_f8_e4m3 = false;
int64_t ne[SD_MAX_DIMS] = {1, 1, 1, 1, 1};
int n_dims = 0;
@@ -59,7 +62,7 @@ struct TensorStorage {
}
int64_t nbytes_to_read() const {
if (is_bf16) {
if (is_bf16 || is_f8_e4m3) {
return nbytes() / 2;
} else {
return nbytes();
@@ -107,6 +110,8 @@ struct TensorStorage {
const char* type_name = ggml_type_name(type);
if (is_bf16) {
type_name = "bf16";
} else if (is_f8_e4m3) {
type_name = "f8_e4m3";
}
ss << name << " | " << type_name << " | ";
ss << n_dims << " [";
@@ -144,6 +149,9 @@ public:
bool init_from_file(const std::string& file_path, const std::string& prefix = "");
SDVersion get_sd_version();
ggml_type get_sd_wtype();
ggml_type get_conditioner_wtype();
ggml_type get_diffusion_model_wtype();
ggml_type get_vae_wtype();
bool load_tensors(on_new_tensor_cb_t on_new_tensor_cb, ggml_backend_t backend);
bool load_tensors(std::map<std::string, struct ggml_tensor*>& tensors,
ggml_backend_t backend,
@@ -155,4 +163,6 @@ public:
static std::string load_merges();
static std::string load_t5_tokenizer_json();
};
#endif // __MODEL_H__
#endif // __MODEL_H__