pull from upstream

CMakeLists.txt

@@ -88,6 +88,10 @@ if (NOT DEFINED GGML_LLAMAFILE)
     set(GGML_LLAMAFILE_DEFAULT ON)
 endif()
 
+if (NOT DEFINED GGML_AMX)
+    set(GGML_AMX ON)
+endif()
+
 if (NOT DEFINED GGML_CUDA_GRAPHS)
     set(GGML_CUDA_GRAPHS_DEFAULT ON)
 endif()

Makefile

@@ -93,11 +93,6 @@ GGML_METAL := 1
 DEPRECATE_WARNING := 1
 endif
 
-ifdef LLAMA_OPENMP
-GGML_OPENMP := 1
-DEPRECATE_WARNING := 1
-endif
-
 ifdef LLAMA_RPC
 GGML_RPC := 1
 DEPRECATE_WARNING := 1
@@ -584,6 +579,11 @@ ifndef GGML_NO_LLAMAFILE
 	OBJ_GGML    += ggml/src/llamafile/sgemm.o
 endif
 
+ifndef GGML_NO_AMX
+	MK_CPPFLAGS += -DGGML_USE_AMX
+	OBJ_GGML    += ggml/src/ggml-amx.o ggml/src/ggml-amx/mmq.o
+endif
+
 ifdef GGML_RPC
 	MK_CPPFLAGS += -DGGML_USE_RPC
 	OBJ_GGML    += ggml/src/ggml-rpc.o
@@ -1087,6 +1087,19 @@ ggml/src/llamafile/sgemm.o: \
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 endif # GGML_NO_LLAMAFILE
 
+ifndef GGML_NO_AMX
+ggml/src/ggml-amx.o: \
+	ggml/src/ggml-amx.cpp \
+	ggml/include/ggml-amx.h
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
+ggml/src/ggml-amx/mmq.o: \
+	ggml/src/ggml-amx/mmq.cpp \
+	ggml/src/ggml-amx/mmq.h \
+	ggml/include/ggml.h
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+endif
+
 ifdef GGML_RPC
 ggml/src/ggml-rpc.o: \
 	ggml/src/ggml-rpc.cpp \
@@ -1238,6 +1251,7 @@ clean:
 	rm -vrf ggml/src/ggml-metal-embed.metal
 	rm -vrf ggml/src/ggml-cuda/*.o
 	rm -vrf ggml/src/ggml-cuda/template-instances/*.o
+	rm -vrf ggml/src/ggml-amx/*.o
 	rm -rvf $(BUILD_TARGETS)
 	rm -rvf $(TEST_TARGETS)
 	rm -f vulkan-shaders-gen ggml/src/ggml-vulkan-shaders.hpp ggml/src/ggml-vulkan-shaders.cpp

README.md

@@ -29,7 +29,7 @@ variety of hardware - locally and in the cloud.
 
 - Plain C/C++ implementation without any dependencies
 - Apple silicon is a first-class citizen - optimized via ARM NEON, Accelerate and Metal frameworks
-- AVX, AVX2 and AVX512 support for x86 architectures
+- AVX, AVX2, AVX512 and AMX support for x86 architectures
 - 1.5-bit, 2-bit, 3-bit, 4-bit, 5-bit, 6-bit, and 8-bit integer quantization for faster inference and reduced memory use
 - Custom CUDA kernels for running LLMs on NVIDIA GPUs (support for AMD GPUs via HIP and Moore Threads MTT GPUs via MUSA)
 - Vulkan and SYCL backend support
@@ -93,6 +93,7 @@ Typically finetunes of the base models below are supported as well.
 - [x] [FalconMamba Models](https://huggingface.co/collections/tiiuae/falconmamba-7b-66b9a580324dd1598b0f6d4a)
 - [x] [Jais](https://huggingface.co/inceptionai/jais-13b-chat)
 - [x] [Bielik-11B-v2.3](https://huggingface.co/collections/speakleash/bielik-11b-v23-66ee813238d9b526a072408a)
+- [x] [RWKV-6](https://github.com/BlinkDL/RWKV-LM)
 
 (instructions for supporting more models: [HOWTO-add-model.md](./docs/development/HOWTO-add-model.md))
 
@@ -122,6 +123,7 @@ Typically finetunes of the base models below are supported as well.
 - Rust (nicer API): [mdrokz/rust-llama.cpp](https://github.com/mdrokz/rust-llama.cpp)
 - Rust (more direct bindings): [utilityai/llama-cpp-rs](https://github.com/utilityai/llama-cpp-rs)
 - C#/.NET: [SciSharp/LLamaSharp](https://github.com/SciSharp/LLamaSharp)
+- C#/VB.NET (more features - community license): [LM-Kit.NET](https://docs.lm-kit.com/lm-kit-net/index.html)
 - Scala 3: [donderom/llm4s](https://github.com/donderom/llm4s)
 - Clojure: [phronmophobic/llama.clj](https://github.com/phronmophobic/llama.clj)
 - React Native: [mybigday/llama.rn](https://github.com/mybigday/llama.rn)
@@ -131,6 +133,7 @@ Typically finetunes of the base models below are supported as well.
 - PHP (API bindings and features built on top of llama.cpp): [distantmagic/resonance](https://github.com/distantmagic/resonance) [(more info)](https://github.com/ggerganov/llama.cpp/pull/6326)
 - Guile Scheme: [guile_llama_cpp](https://savannah.nongnu.org/projects/guile-llama-cpp)
 - Swift [srgtuszy/llama-cpp-swift](https://github.com/srgtuszy/llama-cpp-swift)
+- Swift [ShenghaiWang/SwiftLlama](https://github.com/ShenghaiWang/SwiftLlama)
 
 **UI:**
 
@@ -171,6 +174,7 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 - [LARS - The LLM & Advanced Referencing Solution](https://github.com/abgulati/LARS) (AGPL)
 - [LLMUnity](https://github.com/undreamai/LLMUnity) (MIT)
 - [Llama Assistant](https://github.com/vietanhdev/llama-assistant) (GPL)
+- [PocketPal AI - An iOS and Android App](https://github.com/a-ghorbani/pocketpal-ai) (MIT)
 
 *(to have a project listed here, it should clearly state that it depends on `llama.cpp`)*
 
@@ -186,6 +190,7 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 
 - [Paddler](https://github.com/distantmagic/paddler) - Stateful load balancer custom-tailored for llama.cpp
 - [GPUStack](https://github.com/gpustack/gpustack) - Manage GPU clusters for running LLMs
+- [llama_cpp_canister](https://github.com/onicai/llama_cpp_canister) - llama.cpp as a smart contract on the Internet Computer, using WebAssembly
 
 **Games:**
 - [Lucy's Labyrinth](https://github.com/MorganRO8/Lucys_Labyrinth) - A simple maze game where agents controlled by an AI model will try to trick you.

ci/run.sh

@@ -53,7 +53,7 @@ if [ ! -z ${GG_BUILD_SYCL} ]; then
         exit 1
     fi
 
-    CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_SYCL=1 DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON"
+    CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_SYCL=1 -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON"
 fi
 
 if [ ! -z ${GG_BUILD_VULKAN} ]; then

common/arg.cpp

@@ -947,6 +947,20 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             params.sparams.tfs_z = std::stof(value);
         }
     ).set_sparam());
+    add_opt(common_arg(
+        {"--xtc-probability"}, "N",
+        string_format("xtc probability (default: %.1f, 0.0 = disabled)", (double)params.sparams.xtc_probability),
+        [](common_params & params, const std::string & value) {
+            params.sparams.xtc_probability = std::stof(value);
+        }
+    ).set_sparam());
+    add_opt(common_arg(
+        {"--xtc-threshold"}, "N",
+        string_format("xtc threshold (default: %.1f, 1.0 = disabled)", (double)params.sparams.xtc_threshold),
+        [](common_params & params, const std::string & value) {
+            params.sparams.xtc_threshold = std::stof(value);
+        }
+    ).set_sparam());
     add_opt(common_arg(
         {"--typical"}, "N",
         string_format("locally typical sampling, parameter p (default: %.1f, 1.0 = disabled)", (double)params.sparams.typ_p),
@@ -1083,7 +1097,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
         }
     ).set_examples({LLAMA_EXAMPLE_EMBEDDING, LLAMA_EXAMPLE_RETRIEVAL, LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_POOLING"));
     add_opt(common_arg(
-        {"--attention"}, "{causal,non,causal}",
+        {"--attention"}, "{causal,non-causal}",
         "attention type for embeddings, use model default if unspecified",
         [](common_params & params, const std::string & value) {
             /**/ if (value == "causal") { params.attention_type = LLAMA_ATTENTION_TYPE_CAUSAL; }
@@ -1681,7 +1695,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_examples({LLAMA_EXAMPLE_BENCH}));
     add_opt(common_arg(
         {"--embd-normalize"}, "N",
-        string_format("normalisation for embendings (default: %d) (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)", params.embd_normalize),
+        string_format("normalisation for embeddings (default: %d) (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)", params.embd_normalize),
         [](common_params & params, int value) {
             params.embd_normalize = value;
         }
@@ -1695,7 +1709,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_examples({LLAMA_EXAMPLE_EMBEDDING}));
     add_opt(common_arg(
         {"--embd-separator"}, "STRING",
-        "separator of embendings (default \\n) for example \"<#sep#>\"",
+        "separator of embeddings (default \\n) for example \"<#sep#>\"",
         [](common_params & params, const std::string & value) {
             params.embd_sep = value;
         }

common/common.cpp

@@ -955,7 +955,7 @@ struct common_init_result common_init_from_params(common_params & params) {
         }
 
         if (llama_model_has_encoder(model)) {
-            llama_encode(lctx, llama_batch_get_one(tmp.data(), tmp.size(), 0, 0));
+            llama_encode(lctx, llama_batch_get_one(tmp.data(), tmp.size()));
             llama_token decoder_start_token_id = llama_model_decoder_start_token(model);
             if (decoder_start_token_id == -1) {
                 decoder_start_token_id = bos;
@@ -964,7 +964,7 @@ struct common_init_result common_init_from_params(common_params & params) {
             tmp.push_back(decoder_start_token_id);
         }
         if (llama_model_has_decoder(model)) {
-            llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
+            llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch)));
         }
         llama_kv_cache_clear(lctx);
         llama_synchronize(lctx);
@@ -1035,7 +1035,7 @@ static ggml_type kv_cache_type_from_str(const std::string & s) {
         return GGML_TYPE_Q5_1;
     }
 
-    throw std::runtime_error("Invalid cache type: " + s);
+    throw std::runtime_error("Unsupported cache type: " + s);
 }
 
 struct llama_context_params common_context_params_to_llama(const common_params & params) {
@@ -1047,7 +1047,7 @@ struct llama_context_params common_context_params_to_llama(const common_params &
     cparams.n_ubatch          = params.n_ubatch;
     cparams.n_threads         = params.cpuparams.n_threads;
     cparams.n_threads_batch   = params.cpuparams_batch.n_threads == -1 ?
-                                    params.cpuparams.n_threads : params.cpuparams_batch.n_threads;
+                                params.cpuparams.n_threads : params.cpuparams_batch.n_threads;
     cparams.logits_all        = params.logits_all;
     cparams.embeddings        = params.embedding;
     cparams.rope_scaling_type = params.rope_scaling_type;
@@ -2104,6 +2104,8 @@ void yaml_dump_non_result_info(FILE * stream, const common_params & params, cons
     fprintf(stream, "top_k: %d # default: 40\n", sparams.top_k);
     fprintf(stream, "top_p: %f # default: 0.95\n", sparams.top_p);
     fprintf(stream, "min_p: %f # default: 0.0\n", sparams.min_p);
+    fprintf(stream, "xtc_probability: %f # default: 0.0\n", sparams.xtc_probability);
+    fprintf(stream, "xtc_threshold: %f # default: 0.1\n", sparams.xtc_threshold);
     fprintf(stream, "typ_p: %f # default: 1.0\n", sparams.typ_p);
     fprintf(stream, "verbose_prompt: %s # default: false\n", params.verbose_prompt ? "true" : "false");
     fprintf(stream, "display_prompt: %s # default: true\n", params.display_prompt ? "true" : "false");

common/common.h

@@ -90,6 +90,8 @@ enum common_sampler_type {
     COMMON_SAMPLER_TYPE_TFS_Z       = 4,
     COMMON_SAMPLER_TYPE_TYPICAL_P   = 5,
     COMMON_SAMPLER_TYPE_TEMPERATURE = 6,
+    COMMON_SAMPLER_TYPE_XTC         = 7,
+    COMMON_SAMPLER_TYPE_INFILL      = 8,
 };
 
 // dimensionality reduction methods, used by cvector-generator
@@ -108,6 +110,8 @@ struct common_sampler_params {
     int32_t top_k             = 40;    // <= 0 to use vocab size
     float   top_p             = 0.95f; // 1.0 = disabled
     float   min_p             = 0.05f; // 0.0 = disabled
+    float   xtc_probability   = 0.00f; // 0.0 = disabled
+    float   xtc_threshold     = 0.10f; // > 0.5 disables XTC
     float   tfs_z             = 1.00f; // 1.0 = disabled
     float   typ_p             = 1.00f; // typical_p, 1.0 = disabled
     float   temp              = 0.80f; // <= 0.0 to sample greedily, 0.0 to not output probabilities
@@ -124,13 +128,15 @@ struct common_sampler_params {
     bool    ignore_eos        = false;
     bool    no_perf           = false; // disable performance metrics
 
+
     std::vector<enum common_sampler_type> samplers = {
         COMMON_SAMPLER_TYPE_TOP_K,
         COMMON_SAMPLER_TYPE_TFS_Z,
         COMMON_SAMPLER_TYPE_TYPICAL_P,
         COMMON_SAMPLER_TYPE_TOP_P,
         COMMON_SAMPLER_TYPE_MIN_P,
-        COMMON_SAMPLER_TYPE_TEMPERATURE
+        COMMON_SAMPLER_TYPE_XTC,
+        COMMON_SAMPLER_TYPE_TEMPERATURE,
     };
 
     std::string grammar; // optional BNF-like grammar to constrain sampling
@@ -268,9 +274,9 @@ struct common_params {
 
     // embedding
     bool embedding         = false; // get only sentence embedding
-    int32_t embd_normalize = 2;     // normalisation for embendings (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)
+    int32_t embd_normalize = 2;     // normalisation for embeddings (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)
     std::string embd_out   = "";    // empty = default, "array" = [[],[]...], "json" = openai style, "json+" = same "json" + cosine similarity matrix
-    std::string embd_sep   = "\n";  // separator of embendings
+    std::string embd_sep   = "\n";  // separator of embeddings
     bool reranking         = false; // enable reranking support on server
 
     // server params

common/json-schema-to-grammar.cpp

@@ -611,7 +611,7 @@ class SchemaConverter {
             }
             return join_seq();
         };
-        return _add_rule(name, "\"\\\"\" " + to_rule(transform()) + " \"\\\"\" space");
+        return _add_rule(name, "\"\\\"\" (" + to_rule(transform()) + ") \"\\\"\" space");
     }
 
     /*