sync: update ggml

ggml_extend.hpp

@@ -20,9 +20,9 @@
 #include <unordered_map>
 #include <vector>
 
-#include "ggml/ggml-alloc.h"
-#include "ggml/ggml-backend.h"
-#include "ggml/ggml.h"
+#include "ggml-alloc.h"
+#include "ggml-backend.h"
+#include "ggml.h"
 
 #ifdef SD_USE_CUBLAS
 #include "ggml-cuda.h"

model.cpp

@@ -11,9 +11,9 @@
 #include "util.h"
 #include "vocab.hpp"
 
-#include "ggml/ggml-alloc.h"
-#include "ggml/ggml-backend.h"
-#include "ggml/ggml.h"
+#include "ggml-alloc.h"
+#include "ggml-backend.h"
+#include "ggml.h"
 
 #include "stable-diffusion.h"
 

model.h

@@ -10,8 +10,8 @@
 #include <tuple>
 #include <vector>
 
-#include "ggml/ggml-backend.h"
-#include "ggml/ggml.h"
+#include "ggml-backend.h"
+#include "ggml.h"
 #include "json.hpp"
 #include "zip.h"
 

stable-diffusion.h

@@ -55,10 +55,10 @@ enum schedule_t {
 
 // same as enum ggml_type
 enum sd_type_t {
-    SD_TYPE_F32  = 0,
-    SD_TYPE_F16  = 1,
-    SD_TYPE_Q4_0 = 2,
-    SD_TYPE_Q4_1 = 3,
+    SD_TYPE_F32     = 0,
+    SD_TYPE_F16     = 1,
+    SD_TYPE_Q4_0    = 2,
+    SD_TYPE_Q4_1    = 3,
     // SD_TYPE_Q4_2 = 4, support has been removed
     // SD_TYPE_Q4_3 = 5, support has been removed
     SD_TYPE_Q5_0    = 6,
@@ -86,6 +86,9 @@ enum sd_type_t {
     SD_TYPE_F64     = 28,
     SD_TYPE_IQ1_M   = 29,
     SD_TYPE_BF16    = 30,
+    SD_TYPE_Q4_0_4_4 = 31,
+    SD_TYPE_Q4_0_4_8 = 32,
+    SD_TYPE_Q4_0_8_8 = 33,
     SD_TYPE_COUNT,
 };
 

util.cpp

@@ -22,7 +22,7 @@
 #include <unistd.h>
 #endif
 
-#include "ggml/ggml.h"
+#include "ggml.h"
 #include "stable-diffusion.h"
 
 #define STB_IMAGE_RESIZE_IMPLEMENTATION