llama_cpp 0.12.0 → 0.12.2

data/vendor/tmp/llama.cpp/ggml.h

@@ -187,6 +187,16 @@
 #    define GGML_API
 #endif
 
+#ifdef GGML_MULTIPLATFORM
+#    if defined(_WIN32)
+#        define GGML_CALL
+#    else
+#        define GGML_CALL __attribute__((__ms_abi__))
+#    endif
+#else
+#    define GGML_CALL
+#endif
+
 // TODO: support for clang
 #ifdef __GNUC__
 #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
@@ -218,7 +228,9 @@
 #define GGML_MAX_PARAMS         2048
 #define GGML_MAX_CONTEXTS       64
 #define GGML_MAX_SRC            10
+#ifndef GGML_MAX_NAME
 #define GGML_MAX_NAME           64
+#endif
 #define GGML_MAX_OP_PARAMS      64
 #define GGML_DEFAULT_N_THREADS  4
 #define GGML_DEFAULT_GRAPH_SIZE 2048
@@ -339,6 +351,8 @@ extern "C" {
         GGML_TYPE_Q5_K = 13,
         GGML_TYPE_Q6_K = 14,
         GGML_TYPE_Q8_K = 15,
+        GGML_TYPE_IQ2_XXS = 16,
+        GGML_TYPE_IQ2_XS  = 17,
         GGML_TYPE_I8,
         GGML_TYPE_I16,
         GGML_TYPE_I32,
@@ -373,6 +387,8 @@ extern "C" {
         GGML_FTYPE_MOSTLY_Q4_K = 12, // except 1d tensors
         GGML_FTYPE_MOSTLY_Q5_K = 13, // except 1d tensors
         GGML_FTYPE_MOSTLY_Q6_K = 14, // except 1d tensors
+        GGML_FTYPE_MOSTLY_IQ2_XXS = 15, // except 1d tensors
+        GGML_FTYPE_MOSTLY_IQ2_XS  = 16, // except 1d tensors
     };
 
     // available tensor operations:
@@ -643,41 +659,41 @@ extern "C" {
     GGML_API void    ggml_print_object (const struct ggml_object * obj);
     GGML_API void    ggml_print_objects(const struct ggml_context * ctx);
 
-    GGML_API int64_t ggml_nelements   (const struct ggml_tensor * tensor);
-    GGML_API int64_t ggml_nrows       (const struct ggml_tensor * tensor);
-    GGML_API size_t  ggml_nbytes      (const struct ggml_tensor * tensor);
-    GGML_API size_t  ggml_nbytes_pad  (const struct ggml_tensor * tensor); // same as ggml_nbytes() but padded to GGML_MEM_ALIGN
+    GGML_API GGML_CALL int64_t ggml_nelements   (const struct ggml_tensor * tensor);
+    GGML_API GGML_CALL int64_t ggml_nrows       (const struct ggml_tensor * tensor);
+    GGML_API GGML_CALL size_t  ggml_nbytes      (const struct ggml_tensor * tensor);
+    GGML_API           size_t  ggml_nbytes_pad  (const struct ggml_tensor * tensor); // same as ggml_nbytes() but padded to GGML_MEM_ALIGN
 
-    GGML_API int    ggml_blck_size(enum ggml_type type);
-    GGML_API size_t ggml_type_size(enum ggml_type type);             // size in bytes for all elements in a block
-    GGML_API size_t ggml_row_size (enum ggml_type type, int64_t ne); // size in bytes for all elements in a row
+    GGML_API GGML_CALL int    ggml_blck_size(enum ggml_type type);
+    GGML_API GGML_CALL size_t ggml_type_size(enum ggml_type type);             // size in bytes for all elements in a block
+    GGML_API GGML_CALL size_t ggml_row_size (enum ggml_type type, int64_t ne); // size in bytes for all elements in a row
 
     GGML_DEPRECATED(
     GGML_API double ggml_type_sizef(enum ggml_type type), // ggml_type_size()/ggml_blck_size() as float
     "use ggml_row_size() instead");
 
-    GGML_API const char * ggml_type_name(enum ggml_type type);
-    GGML_API const char * ggml_op_name  (enum ggml_op   op);
-    GGML_API const char * ggml_op_symbol(enum ggml_op   op);
+    GGML_API GGML_CALL const char * ggml_type_name(enum ggml_type type);
+    GGML_API GGML_CALL const char * ggml_op_name  (enum ggml_op   op);
+    GGML_API           const char * ggml_op_symbol(enum ggml_op   op);
 
-    GGML_API const char * ggml_unary_op_name(enum ggml_unary_op op);
-    GGML_API const char * ggml_op_desc(const struct ggml_tensor * t); // unary or op name
+    GGML_API           const char * ggml_unary_op_name(enum ggml_unary_op op);
+    GGML_API GGML_CALL const char * ggml_op_desc(const struct ggml_tensor * t); // unary or op name
 
-    GGML_API size_t  ggml_element_size(const struct ggml_tensor * tensor);
+    GGML_API GGML_CALL size_t  ggml_element_size(const struct ggml_tensor * tensor);
 
-    GGML_API bool    ggml_is_quantized(enum ggml_type type);
+    GGML_API GGML_CALL bool    ggml_is_quantized(enum ggml_type type);
 
     // TODO: temporary until model loading of ggml examples is refactored
     GGML_API enum ggml_type ggml_ftype_to_ggml_type(enum ggml_ftype ftype);
 
-    GGML_API bool ggml_is_transposed(const struct ggml_tensor * tensor);
-    GGML_API bool ggml_is_contiguous(const struct ggml_tensor * tensor);
-    GGML_API bool ggml_is_permuted  (const struct ggml_tensor * tensor);
-    GGML_API bool ggml_is_scalar    (const struct ggml_tensor * tensor);
-    GGML_API bool ggml_is_vector    (const struct ggml_tensor * tensor);
-    GGML_API bool ggml_is_matrix    (const struct ggml_tensor * tensor);
-    GGML_API bool ggml_is_3d        (const struct ggml_tensor * tensor);
-    GGML_API int  ggml_n_dims       (const struct ggml_tensor * tensor); // returns 1 for scalars
+    GGML_API GGML_CALL bool ggml_is_transposed(const struct ggml_tensor * tensor);
+    GGML_API GGML_CALL bool ggml_is_contiguous(const struct ggml_tensor * tensor);
+    GGML_API GGML_CALL bool ggml_is_permuted  (const struct ggml_tensor * tensor);
+    GGML_API           bool ggml_is_scalar    (const struct ggml_tensor * tensor);
+    GGML_API           bool ggml_is_vector    (const struct ggml_tensor * tensor);
+    GGML_API           bool ggml_is_matrix    (const struct ggml_tensor * tensor);
+    GGML_API           bool ggml_is_3d        (const struct ggml_tensor * tensor);
+    GGML_API           int  ggml_n_dims       (const struct ggml_tensor * tensor); // returns 1 for scalars
 
     GGML_API bool ggml_are_same_shape(const struct ggml_tensor * t0, const struct ggml_tensor * t1);
 
@@ -764,7 +780,7 @@ extern "C" {
     GGML_API void *  ggml_get_data    (const struct ggml_tensor * tensor);
     GGML_API float * ggml_get_data_f32(const struct ggml_tensor * tensor);
 
-    GGML_API enum ggml_unary_op ggml_get_unary_op(const struct ggml_tensor * tensor);
+    GGML_API GGML_CALL enum ggml_unary_op ggml_get_unary_op(const struct ggml_tensor * tensor);
 
     GGML_API const char *         ggml_get_name   (const struct ggml_tensor * tensor);
     GGML_API struct ggml_tensor * ggml_set_name   (      struct ggml_tensor * tensor, const char * name);
@@ -1159,22 +1175,16 @@ extern "C" {
             struct ggml_tensor  * a,
             struct ggml_tensor  * b);
 
-    // a -> b, in-place, return view(b)
-    GGML_API struct ggml_tensor * ggml_cpy_inplace(
+    GGML_API struct ggml_tensor * ggml_cast(
             struct ggml_context * ctx,
             struct ggml_tensor  * a,
-            struct ggml_tensor  * b);
+            enum   ggml_type      type);
 
     // make contiguous
     GGML_API struct ggml_tensor * ggml_cont(
             struct ggml_context * ctx,
             struct ggml_tensor  * a);
 
-    // make contiguous, in-place
-    GGML_API struct ggml_tensor * ggml_cont_inplace(
-            struct ggml_context * ctx,
-            struct ggml_tensor  * a);
-
     // make contiguous, with new shape
     GGML_API struct ggml_tensor * ggml_cont_1d(
             struct ggml_context * ctx,
@@ -1413,7 +1423,7 @@ extern "C" {
             float                 beta_slow);
 
     // compute correction dims for YaRN RoPE scaling
-    void ggml_rope_yarn_corr_dims(
+    GGML_CALL void ggml_rope_yarn_corr_dims(
         int n_dims, int n_orig_ctx, float freq_base, float beta_fast, float beta_slow, float dims[2]);
 
     // xPos RoPE, in-place, returns view(a)
@@ -1847,8 +1857,8 @@ extern "C" {
 
     // ggml_graph_plan() has to be called before ggml_graph_compute()
     // when plan.work_size > 0, caller must allocate memory for plan.work_data
-    GGML_API struct ggml_cplan ggml_graph_plan   (struct ggml_cgraph * cgraph, int n_threads /*= GGML_DEFAULT_N_THREADS*/);
-    GGML_API int               ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cplan * cplan);
+    GGML_API struct ggml_cplan ggml_graph_plan   (const struct ggml_cgraph * cgraph, int n_threads /*= GGML_DEFAULT_N_THREADS*/);
+    GGML_API int               ggml_graph_compute(      struct ggml_cgraph * cgraph, struct ggml_cplan * cplan);
 
     // same as ggml_graph_compute() but the work data is allocated as a part of the context
     // note: the drawback of this API is that you must have ensured that the context has enough memory for the work data
@@ -2068,7 +2078,18 @@ extern "C" {
     GGML_API size_t ggml_quantize_q5_K(const float * src, void * dst, int n, int k, int64_t * hist);
     GGML_API size_t ggml_quantize_q6_K(const float * src, void * dst, int n, int k, int64_t * hist);
 
-    GGML_API size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, int start, int n, int64_t * hist);
+    GGML_API size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst,
+            int start, int nrows, int n_per_row, int64_t * hist, const float * imatrix);
+
+    // These are needed for IQ2_XS and IQ2_XXS quantizations
+    GGML_API void ggml_init_iq2_quantization(enum ggml_type type);
+    GGML_API void ggml_deinit_iq2_quantization(enum ggml_type type);
+
+    //
+    // Importance matrix
+    //
+    typedef void(*ggml_collect_imatrix_t)(const struct ggml_tensor * src0, const struct ggml_tensor * src1);
+    GGML_API void ggml_set_imatrix_collection(ggml_collect_imatrix_t imatrix_collect);
 
     //
     // gguf