Merge branch 'master' into xsn/vision_2

.github/workflows/build.yml

@@ -87,6 +87,7 @@ jobs:
         if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
         run: |
           cp LICENSE ./build/bin/
+          cp examples/run/linenoise.cpp/LICENSE ./build/bin/LICENSE.linenoise.cpp
           zip -r llama-${{ steps.tag.outputs.name }}-bin-macos-arm64.zip ./build/bin/*
 
       - name: Upload artifacts
@@ -149,6 +150,7 @@ jobs:
         if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
         run: |
           cp LICENSE ./build/bin/
+          cp examples/run/linenoise.cpp/LICENSE ./build/bin/LICENSE.linenoise.cpp
           zip -r llama-${{ steps.tag.outputs.name }}-bin-macos-x64.zip ./build/bin/*
 
       - name: Upload artifacts
@@ -217,6 +219,7 @@ jobs:
         if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
         run: |
           cp LICENSE ./build/bin/
+          cp examples/run/linenoise.cpp/LICENSE ./build/bin/LICENSE.linenoise.cpp
           zip -r llama-${{ steps.tag.outputs.name }}-bin-ubuntu-x64.zip ./build/bin/*
 
       - name: Upload artifacts
@@ -234,7 +237,7 @@ jobs:
     strategy:
       matrix:
         sanitizer: [ADDRESS, THREAD, UNDEFINED]
-        build_type: [Debug, Release]
+        build_type: [Debug]
 
     steps:
       - name: Clone
@@ -796,6 +799,7 @@ jobs:
         if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
         run: |
           Copy-Item LICENSE .\build\bin\Release\llama.cpp.txt
+          Copy-Item .\examples\run\linenoise.cpp\LICENSE .\build\bin\Release\linenoise.cpp.txt
           7z a llama-${{ steps.tag.outputs.name }}-bin-win-${{ matrix.build }}.zip .\build\bin\Release\*
 
       - name: Upload artifacts

.github/workflows/server.yml

@@ -112,9 +112,9 @@ jobs:
               -DGGML_OPENMP=OFF ;
           cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
 
-      - name: Build
-        id: cmake_build
-        if: ${{ matrix.sanitizer != 'THREAD' }}
+      - name: Build (sanitizers)
+        id: cmake_build_sanitizers
+        if: ${{ matrix.sanitizer != '' && matrix.sanitizer != 'THREAD' }}
         run: |
           cmake -B build \
               -DGGML_NATIVE=OFF \
@@ -124,12 +124,31 @@ jobs:
               -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ;
           cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
 
+      - name: Build (sanitizers)
+        id: cmake_build
+        if: ${{ matrix.sanitizer == '' }}
+        run: |
+          cmake -B build \
+              -DGGML_NATIVE=OFF \
+              -DLLAMA_BUILD_SERVER=ON \
+              -DLLAMA_CURL=ON \
+              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} ;
+          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
+
       - name: Tests
         id: server_integration_tests
+        if: ${{ matrix.sanitizer == '' }}
         run: |
           cd examples/server/tests
           ./tests.sh
 
+      - name: Tests (sanitizers)
+        id: server_integration_tests_sanitizers
+        if: ${{ matrix.sanitizer != '' }}
+        run: |
+          cd examples/server/tests
+          LLAMA_SANITIZE=1 ./tests.sh
+
       - name: Slow tests
         id: server_integration_tests_slow
         if: ${{ (github.event.schedule || github.event.inputs.slow_tests == 'true') && matrix.build_type == 'Release' }}

CMakeLists.txt

@@ -83,11 +83,8 @@ include(${CMAKE_CURRENT_SOURCE_DIR}/cmake/build-info.cmake)
 include(${CMAKE_CURRENT_SOURCE_DIR}/cmake/common.cmake)
 
 # override ggml options
-set(GGML_SANITIZE_THREAD    ${LLAMA_SANITIZE_THREAD})
-set(GGML_SANITIZE_ADDRESS   ${LLAMA_SANITIZE_ADDRESS})
-set(GGML_SANITIZE_UNDEFINED ${LLAMA_SANITIZE_UNDEFINED})
-set(GGML_ALL_WARNINGS       ${LLAMA_ALL_WARNINGS})
-set(GGML_FATAL_WARNINGS     ${LLAMA_FATAL_WARNINGS})
+set(GGML_ALL_WARNINGS   ${LLAMA_ALL_WARNINGS})
+set(GGML_FATAL_WARNINGS ${LLAMA_FATAL_WARNINGS})
 
 # change the default for these ggml options
 if (NOT DEFINED GGML_LLAMAFILE)
@@ -117,16 +114,62 @@ llama_option_depr(WARNING     LLAMA_SYCL                GGML_SYCL)
 llama_option_depr(WARNING     LLAMA_SYCL_F16            GGML_SYCL_F16)
 llama_option_depr(WARNING     LLAMA_CANN                GGML_CANN)
 
+if (NOT MSVC)
+    if (LLAMA_SANITIZE_THREAD)
+        message(STATUS "Using -fsanitize=thread")
+
+        add_compile_options(-fsanitize=thread)
+        link_libraries     (-fsanitize=thread)
+    endif()
+
+    if (LLAMA_SANITIZE_ADDRESS)
+        message(STATUS "Using -fsanitize=address")
+
+        add_compile_options(-fsanitize=address -fno-omit-frame-pointer)
+        link_libraries     (-fsanitize=address)
+    endif()
+
+    if (LLAMA_SANITIZE_UNDEFINED)
+        message(STATUS "Using -fsanitize=undefined")
+
+        add_compile_options(-fsanitize=undefined)
+        link_libraries     (-fsanitize=undefined)
+    endif()
+endif()
+
 #
-# build the library
+# 3rd-party
 #
 
 if (NOT TARGET ggml)
     add_subdirectory(ggml)
     # ... otherwise assume ggml is added by a parent CMakeLists.txt
 endif()
+
+#
+# build the library
+#
+
 add_subdirectory(src)
 
+#
+# utils, programs, examples and tests
+#
+
+if (LLAMA_BUILD_COMMON)
+    add_subdirectory(common)
+endif()
+
+if (LLAMA_BUILD_COMMON AND LLAMA_BUILD_TESTS AND NOT CMAKE_JS_VERSION)
+    include(CTest)
+    add_subdirectory(tests)
+endif()
+
+if (LLAMA_BUILD_COMMON AND LLAMA_BUILD_EXAMPLES)
+    add_subdirectory(examples)
+    add_subdirectory(pocs)
+endif()
+
 #
 # install
 #
@@ -200,21 +243,3 @@ configure_file(cmake/llama.pc.in
 
 install(FILES "${CMAKE_CURRENT_BINARY_DIR}/llama.pc"
         DESTINATION lib/pkgconfig)
-
-#
-# utils, programs, examples and tests
-#
-
-if (LLAMA_BUILD_COMMON)
-    add_subdirectory(common)
-endif()
-
-if (LLAMA_BUILD_COMMON AND LLAMA_BUILD_TESTS AND NOT CMAKE_JS_VERSION)
-    include(CTest)
-    add_subdirectory(tests)
-endif()
-
-if (LLAMA_BUILD_COMMON AND LLAMA_BUILD_EXAMPLES)
-    add_subdirectory(examples)
-    add_subdirectory(pocs)
-endif()

Makefile

@@ -1361,7 +1361,9 @@ llama-server: \
 	examples/server/httplib.h \
 	examples/server/index.html.hpp \
 	examples/server/loading.html.hpp \
+	common/chat-template.hpp \
 	common/json.hpp \
+	common/minja.hpp \
 	$(OBJ_ALL)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h %.hpp $<,$^) -Iexamples/server $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS) $(LWINSOCK2)

cmake/build-info.cmake

@@ -44,7 +44,7 @@ if(MSVC)
     set(BUILD_TARGET ${CMAKE_VS_PLATFORM_NAME})
 else()
     execute_process(
-        COMMAND sh -c "$@ --version | head -1" _ ${CMAKE_C_COMPILER}
+        COMMAND sh -c "\"$@\" --version | head -1" _ ${CMAKE_C_COMPILER}
         OUTPUT_VARIABLE OUT
         OUTPUT_STRIP_TRAILING_WHITESPACE
     )

common/CMakeLists.txt

@@ -56,6 +56,7 @@ add_library(${TARGET} STATIC
     arg.cpp
     arg.h
     base64.hpp
+    chat-template.hpp
     common.cpp
     common.h
     console.cpp
@@ -64,6 +65,7 @@ add_library(${TARGET} STATIC
     json.hpp
     log.cpp
     log.h
+    minja.hpp
     ngram-cache.cpp
     ngram-cache.h
     sampling.cpp

common/arg.cpp

@@ -133,7 +133,8 @@ static void common_params_handle_model_default(
         const std::string & model_url,
         std::string & hf_repo,
         std::string & hf_file,
-        const std::string & hf_token) {
+        const std::string & hf_token,
+        const std::string & model_default) {
     if (!hf_repo.empty()) {
         // short-hand to avoid specifying --hf-file -> default it to --model
         if (hf_file.empty()) {
@@ -163,7 +164,7 @@ static void common_params_handle_model_default(
             model = fs_get_cache_file(string_split<std::string>(f, '/').back());
         }
     } else if (model.empty()) {
-        model = DEFAULT_MODEL_PATH;
+        model = model_default;
     }
 }
 
@@ -299,8 +300,9 @@ static bool common_params_parse_ex(int argc, char ** argv, common_params_context
     }
 
     // TODO: refactor model params in a common struct
-    common_params_handle_model_default(params.model,         params.model_url,         params.hf_repo,         params.hf_file,         params.hf_token);
-    common_params_handle_model_default(params.vocoder.model, params.vocoder.model_url, params.vocoder.hf_repo, params.vocoder.hf_file, params.hf_token);
+    common_params_handle_model_default(params.model,             params.model_url,             params.hf_repo,             params.hf_file,             params.hf_token, DEFAULT_MODEL_PATH);
+    common_params_handle_model_default(params.speculative.model, params.speculative.model_url, params.speculative.hf_repo, params.speculative.hf_file, params.hf_token, "");
+    common_params_handle_model_default(params.vocoder.model,     params.vocoder.model_url,     params.vocoder.hf_repo,     params.vocoder.hf_file,     params.hf_token, "");
 
     if (params.escape) {
         string_process_escapes(params.prompt);
@@ -323,6 +325,14 @@ static bool common_params_parse_ex(int argc, char ** argv, common_params_context
         throw std::invalid_argument("error: either --embedding or --reranking can be specified, but not both");
     }
 
+    if (!params.chat_template.empty() && !common_chat_verify_template(params.chat_template, params.use_jinja)) {
+        throw std::runtime_error(string_format(
+            "error: the supplied chat template is not supported: %s%s\n",
+            params.chat_template.c_str(),
+            params.use_jinja ? "" : "\nnote: llama.cpp was started without --jinja, we only support commonly used templates"
+        ));
+    }
+
     return true;
 }
 
@@ -1629,6 +1639,13 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             params.hf_repo = value;
         }
     ).set_env("LLAMA_ARG_HF_REPO"));
+    add_opt(common_arg(
+        {"-hfd", "-hfrd", "--hf-repo-draft"}, "<user>/<model>[:quant]",
+        "Same as --hf-repo, but for the draft model (default: unused)",
+        [](common_params & params, const std::string & value) {
+            params.speculative.hf_repo = value;
+        }
+    ).set_env("LLAMA_ARG_HFD_REPO"));
     add_opt(common_arg(
         {"-hff", "--hf-file"}, "FILE",
         "Hugging Face model file. If specified, it will override the quant in --hf-repo (default: unused)",
@@ -1938,24 +1955,44 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             }
         }
     ).set_examples({LLAMA_EXAMPLE_SERVER}));
+    add_opt(common_arg(
+        {"--jinja"},
+        "use jinja template for chat (default: disabled)",
+        [](common_params & params) {
+            params.use_jinja = true;
+        }
+    ).set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_MAIN}).set_env("LLAMA_ARG_JINJA"));
     add_opt(common_arg(
         {"--chat-template"}, "JINJA_TEMPLATE",
         string_format(
             "set custom jinja chat template (default: template taken from model's metadata)\n"
             "if suffix/prefix are specified, template will be disabled\n"
+            "only commonly used templates are accepted (unless --jinja is set before this flag):\n"
             "list of built-in templates:\n%s", list_builtin_chat_templates().c_str()
         ),
         [](common_params & params, const std::string & value) {
-            if (!common_chat_verify_template(value)) {
-                throw std::runtime_error(string_format(
-                    "error: the supplied chat template is not supported: %s\n"
-                    "note: llama.cpp does not use jinja parser, we only support commonly used templates\n",
-                    value.c_str()
-                ));
-            }
             params.chat_template = value;
         }
     ).set_examples({LLAMA_EXAMPLE_MAIN, LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CHAT_TEMPLATE"));
+    add_opt(common_arg(
+        {"--chat-template-file"}, "JINJA_TEMPLATE_FILE",
+        string_format(
+            "set custom jinja chat template file (default: template taken from model's metadata)\n"
+            "if suffix/prefix are specified, template will be disabled\n"
+            "only commonly used templates are accepted (unless --jinja is set before this flag):\n"
+            "list of built-in templates:\n%s", list_builtin_chat_templates().c_str()
+        ),
+        [](common_params & params, const std::string & value) {
+            std::ifstream file(value);
+            if (!file) {
+                throw std::runtime_error(string_format("error: failed to open file '%s'\n", value.c_str()));
+            }
+            std::copy(
+                std::istreambuf_iterator<char>(file),
+                std::istreambuf_iterator<char>(),
+                std::back_inserter(params.chat_template));
+        }
+    ).set_examples({LLAMA_EXAMPLE_MAIN, LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CHAT_TEMPLATE_FILE"));
     add_opt(common_arg(
         {"-sps", "--slot-prompt-similarity"}, "SIMILARITY",
         string_format("how much the prompt of a request must match the prompt of a slot in order to use that slot (default: %.2f, 0.0 = disabled)\n", params.slot_prompt_similarity),
@@ -2254,6 +2291,13 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             params.vocoder.model = value;
         }
     ).set_examples({LLAMA_EXAMPLE_TTS, LLAMA_EXAMPLE_SERVER}));
+     add_opt(common_arg(
+        {"--tts-use-guide-tokens"},
+        "Use guide tokens to improve TTS word recall",
+        [](common_params & params) {
+            params.vocoder.use_guide_tokens = true;
+        }
+    ).set_examples({LLAMA_EXAMPLE_TTS, LLAMA_EXAMPLE_SERVER}));
 
     // model-specific
     add_opt(common_arg(