cui-llama.rn 1.5.0 → 1.6.1

package/android/build.gradle

@@ -1,116 +1,116 @@
-buildscript {
-  repositories {
-    google()
-    mavenCentral()
-  }
-
-  dependencies {
-    classpath "com.android.tools.build:gradle:7.2.1"
-  }
-}
-
-def isNewArchitectureEnabled() {
-  return rootProject.hasProperty("newArchEnabled") && rootProject.getProperty("newArchEnabled") == "true"
-}
-
-apply plugin: "com.android.library"
-
-
-def appProject = rootProject.allprojects.find { it.plugins.hasPlugin('com.android.application') }
-
-if (isNewArchitectureEnabled()) {
-  apply plugin: "com.facebook.react"
-}
-
-def getExtOrDefault(name) {
-  return rootProject.ext.has(name) ? rootProject.ext.get(name) : project.properties["RNLlama_" + name]
-}
-
-def getExtOrIntegerDefault(name) {
-  return rootProject.ext.has(name) ? rootProject.ext.get(name) : (project.properties["RNLlama_" + name]).toInteger()
-}
-
-def reactNativeArchitectures() {
-  def value = project.getProperties().get("reactNativeArchitectures")
-  def archs = value ? value.split(",") : ["x86_64", "arm64-v8a"]
-  return archs.findAll { it != "armeabi-v7a" && it != "x86" } // Not building for 32-bit architectures
-}
-
-android {
-  ndkVersion getExtOrDefault("ndkVersion")
-  def ndkVersionMajor = ndkVersion.split("\\.")[0].toInteger()
-  if (ndkVersionMajor < 24) {
-    ndkVersion = project.properties["RNLlama_ndkversion"]
-  }
-  compileSdkVersion getExtOrIntegerDefault("compileSdkVersion")
-
-  defaultConfig {
-    minSdkVersion getExtOrIntegerDefault("minSdkVersion")
-    targetSdkVersion getExtOrIntegerDefault("targetSdkVersion")
-    buildConfigField "boolean", "IS_NEW_ARCHITECTURE_ENABLED", isNewArchitectureEnabled().toString()
-    externalNativeBuild {
-      cmake {
-        abiFilters (*reactNativeArchitectures())
-      }
-    }
-  }
-  def rnllamaBuildFromSource = project.properties["rnllamaBuildFromSource"]
-  if (rnllamaBuildFromSource == "true") {
-    externalNativeBuild {
-      cmake {
-        path = file('src/main/CMakeLists.txt')
-      }
-    }
-    // Exclude jniLibs
-    sourceSets {
-      main {
-        jniLibs.srcDirs = []
-      }
-    }
-  }
-  buildTypes {
-    release {
-      minifyEnabled false
-    }
-  }
-
-  lintOptions {
-    disable "GradleCompatible"
-  }
-
-  compileOptions {
-    sourceCompatibility JavaVersion.VERSION_1_8
-    targetCompatibility JavaVersion.VERSION_1_8
-  }
-
-  sourceSets {
-    main {
-      if (isNewArchitectureEnabled()) {
-        java.srcDirs += ['src/newarch']
-      } else {
-        java.srcDirs += ['src/oldarch']
-      }
-    }
-  }
-}
-
-repositories {
-  mavenCentral()
-  google()
-}
-
-
-dependencies {
-  // For < 0.71, this will be from the local maven repo
-  // For > 0.71, this will be replaced by `com.facebook.react:react-android:$version` by react gradle plugin
-  //noinspection GradleDynamicVersion
-  implementation "com.facebook.react:react-native:+"
-}
-
-if (isNewArchitectureEnabled()) {
-  react {
-    jsRootDir = file("../src/")
-    libraryName = "RNLlama"
-    codegenJavaPackageName = "com.rnllama"
-  }
-}
+buildscript {
+  repositories {
+    google()
+    mavenCentral()
+  }
+
+  dependencies {
+    classpath "com.android.tools.build:gradle:7.2.1"
+  }
+}
+
+def isNewArchitectureEnabled() {
+  return rootProject.hasProperty("newArchEnabled") && rootProject.getProperty("newArchEnabled") == "true"
+}
+
+apply plugin: "com.android.library"
+
+
+def appProject = rootProject.allprojects.find { it.plugins.hasPlugin('com.android.application') }
+
+if (isNewArchitectureEnabled()) {
+  apply plugin: "com.facebook.react"
+}
+
+def getExtOrDefault(name) {
+  return rootProject.ext.has(name) ? rootProject.ext.get(name) : project.properties["RNLlama_" + name]
+}
+
+def getExtOrIntegerDefault(name) {
+  return rootProject.ext.has(name) ? rootProject.ext.get(name) : (project.properties["RNLlama_" + name]).toInteger()
+}
+
+def reactNativeArchitectures() {
+  def value = project.getProperties().get("reactNativeArchitectures")
+  def archs = value ? value.split(",") : ["x86_64", "arm64-v8a"]
+  return archs.findAll { it != "armeabi-v7a" && it != "x86" } // Not building for 32-bit architectures
+}
+
+android {
+  ndkVersion getExtOrDefault("ndkVersion")
+  def ndkVersionMajor = ndkVersion.split("\\.")[0].toInteger()
+  if (ndkVersionMajor < 24) {
+    ndkVersion = project.properties["RNLlama_ndkversion"]
+  }
+  compileSdkVersion getExtOrIntegerDefault("compileSdkVersion")
+
+  defaultConfig {
+    minSdkVersion getExtOrIntegerDefault("minSdkVersion")
+    targetSdkVersion getExtOrIntegerDefault("targetSdkVersion")
+    buildConfigField "boolean", "IS_NEW_ARCHITECTURE_ENABLED", isNewArchitectureEnabled().toString()
+    externalNativeBuild {
+      cmake {
+        abiFilters (*reactNativeArchitectures())
+      }
+    }
+  }
+  def rnllamaBuildFromSource = project.properties["rnllamaBuildFromSource"]
+  if (rnllamaBuildFromSource == "true") {
+    externalNativeBuild {
+      cmake {
+        path = file('src/main/CMakeLists.txt')
+      }
+    }
+    // Exclude jniLibs
+    sourceSets {
+      main {
+        jniLibs.srcDirs = []
+      }
+    }
+  }
+  buildTypes {
+    release {
+      minifyEnabled false
+    }
+  }
+
+  lintOptions {
+    disable "GradleCompatible"
+  }
+
+  compileOptions {
+    sourceCompatibility JavaVersion.VERSION_1_8
+    targetCompatibility JavaVersion.VERSION_1_8
+  }
+
+  sourceSets {
+    main {
+      if (isNewArchitectureEnabled()) {
+        java.srcDirs += ['src/newarch']
+      } else {
+        java.srcDirs += ['src/oldarch']
+      }
+    }
+  }
+}
+
+repositories {
+  mavenCentral()
+  google()
+}
+
+
+dependencies {
+  // For < 0.71, this will be from the local maven repo
+  // For > 0.71, this will be replaced by `com.facebook.react:react-android:$version` by react gradle plugin
+  //noinspection GradleDynamicVersion
+  implementation "com.facebook.react:react-native:+"
+}
+
+if (isNewArchitectureEnabled()) {
+  react {
+    jsRootDir = file("../src/")
+    libraryName = "RNLlama"
+    codegenJavaPackageName = "com.rnllama"
+  }
+}

package/android/gradle.properties

@@ -1,5 +1,5 @@
-RNLlama_kotlinVersion=1.7.0
-RNLlama_minSdkVersion=23
-RNLlama_targetSdkVersion=31
-RNLlama_compileSdkVersion=31
-RNLlama_ndkversion=25.1.8937393
+RNLlama_kotlinVersion=1.7.0
+RNLlama_minSdkVersion=23
+RNLlama_targetSdkVersion=31
+RNLlama_compileSdkVersion=31
+RNLlama_ndkversion=25.1.8937393

package/android/src/main/AndroidManifest.xml

@@ -1,4 +1,4 @@
-<manifest xmlns:android="http://schemas.android.com/apk/res/android"
-          package="com.rnllama">
-
-</manifest>
+<manifest xmlns:android="http://schemas.android.com/apk/res/android"
+          package="com.rnllama">
+
+</manifest>

package/android/src/main/CMakeLists.txt

@@ -1,124 +1,129 @@
-cmake_minimum_required(VERSION 3.10)
-
-project(llama.rn)
-
-find_program(CCACHE_FOUND ccache)
-if(CCACHE_FOUND)
-  set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ccache)
-  set_property(GLOBAL PROPERTY RULE_LAUNCH_LINK ccache)
-endif(CCACHE_FOUND)
-
-set(CMAKE_CXX_STANDARD 17)
-set(RNLLAMA_LIB_DIR ${CMAKE_SOURCE_DIR}/../../../cpp)
-
-include_directories(${RNLLAMA_LIB_DIR})
-
-set(
-    SOURCE_FILES
-    ${RNLLAMA_LIB_DIR}/ggml.c
-    ${RNLLAMA_LIB_DIR}/ggml-alloc.c
-    ${RNLLAMA_LIB_DIR}/ggml-backend.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-backend-reg.cpp
-    ${RNLLAMA_LIB_DIR}/ops.cpp
-    ${RNLLAMA_LIB_DIR}/unary-ops.cpp
-    ${RNLLAMA_LIB_DIR}/binary-ops.cpp
-    ${RNLLAMA_LIB_DIR}/vec.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-cpu.c
-    ${RNLLAMA_LIB_DIR}/ggml-cpu.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-cpu-aarch64.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-cpu-quants.c
-    ${RNLLAMA_LIB_DIR}/ggml-cpu-traits.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-opt.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-threading.cpp
-    ${RNLLAMA_LIB_DIR}/ggml-quants.c
-    ${RNLLAMA_LIB_DIR}/gguf.cpp
-    ${RNLLAMA_LIB_DIR}/log.cpp
-    ${RNLLAMA_LIB_DIR}/llama-impl.cpp
-    ${RNLLAMA_LIB_DIR}/llama-grammar.cpp
-    ${RNLLAMA_LIB_DIR}/llama-sampling.cpp
-    ${RNLLAMA_LIB_DIR}/llama-vocab.cpp
-    ${RNLLAMA_LIB_DIR}/llama-adapter.cpp
-    ${RNLLAMA_LIB_DIR}/llama-chat.cpp
-    ${RNLLAMA_LIB_DIR}/llama-context.cpp
-    ${RNLLAMA_LIB_DIR}/llama-kv-cache.cpp
-    ${RNLLAMA_LIB_DIR}/llama-arch.cpp
-    ${RNLLAMA_LIB_DIR}/llama-batch.cpp
-    ${RNLLAMA_LIB_DIR}/llama-cparams.cpp
-    ${RNLLAMA_LIB_DIR}/llama-hparams.cpp
-    ${RNLLAMA_LIB_DIR}/llama.cpp
-    ${RNLLAMA_LIB_DIR}/llama-model.cpp
-    ${RNLLAMA_LIB_DIR}/llama-model-loader.cpp
-    ${RNLLAMA_LIB_DIR}/llama-mmap.cpp
-    ${RNLLAMA_LIB_DIR}/llama-vocab.cpp
-    ${RNLLAMA_LIB_DIR}/llama-memory.cpp
-    ${RNLLAMA_LIB_DIR}/llama-io.cpp
-    ${RNLLAMA_LIB_DIR}/llama-graph.cpp
-    ${RNLLAMA_LIB_DIR}/sampling.cpp
-    ${RNLLAMA_LIB_DIR}/unicode-data.cpp
-    ${RNLLAMA_LIB_DIR}/unicode.cpp
-    ${RNLLAMA_LIB_DIR}/sgemm.cpp
-    ${RNLLAMA_LIB_DIR}/common.cpp
-    ${RNLLAMA_LIB_DIR}/chat.cpp
-    ${RNLLAMA_LIB_DIR}/json-schema-to-grammar.cpp
-    ${RNLLAMA_LIB_DIR}/json.hpp
-    ${RNLLAMA_LIB_DIR}/minja/minja.hpp
-    ${RNLLAMA_LIB_DIR}/minja/chat-template.hpp
-    ${RNLLAMA_LIB_DIR}/rn-llama.cpp
-    ${CMAKE_SOURCE_DIR}/jni-utils.h
-    ${CMAKE_SOURCE_DIR}/jni.cpp
-)
-
-find_library(LOG_LIB log)
-
-function(build_library target_name cpu_flags)
-    add_library(
-        ${target_name}
-        SHARED
-        ${SOURCE_FILES}
-    )
-
-    target_link_libraries(${target_name} ${LOG_LIB} android)
-
-    target_compile_options(${target_name} PRIVATE -DLM_GGML_USE_CPU -DLM_GGML_USE_CPU_AARCH64 -pthread ${cpu_flags})
-
-    if (${CMAKE_BUILD_TYPE} STREQUAL "Debug")
-        target_compile_options(${target_name} PRIVATE -DRNLLAMA_ANDROID_ENABLE_LOGGING)
-    endif ()
-
-    # NOTE: If you want to debug the native code, you can uncomment if and endif
-    # Note that it will be extremely slow
-    # if (NOT ${CMAKE_BUILD_TYPE} STREQUAL "Debug")
-        target_compile_options(${target_name} PRIVATE -O3 -DNDEBUG -DRNLLAMA_USE_FD_FILE)
-        target_compile_options(${target_name} PRIVATE -fvisibility=hidden -fvisibility-inlines-hidden)
-        target_compile_options(${target_name} PRIVATE -ffunction-sections -fdata-sections)
-
-        target_link_options(${target_name} PRIVATE -Wl,--gc-sections)
-        target_link_options(${target_name} PRIVATE -Wl,--exclude-libs,ALL)
-        target_link_options(${target_name} PRIVATE -flto)
-    # endif ()
-endfunction()
-
-
-# Default target (no specific CPU features)
-build_library("rnllama" "")
-
-if (${ANDROID_ABI} STREQUAL "arm64-v8a")
-    # ARM64 targets
-    # Removing fp16 for now as it leads to issues with some models like deepseek r1 distills
-    # https://github.com/mybigday/llama.rn/pull/110#issuecomment-2609918310
-    build_library("rnllama_v8" "-march=armv8-a")
-    build_library("rnllama_v8_2" "-march=armv8.2-a")
-    build_library("rnllama_v8_2_dotprod" "-march=armv8.2-a+dotprod")
-    build_library("rnllama_v8_2_i8mm" "-march=armv8.2-a+i8mm")
-    build_library("rnllama_v8_2_dotprod_i8mm" "-march=armv8.2-a+dotprod+i8mm")
-
-    # https://github.com/ggerganov/llama.cpp/blob/master/docs/android.md#cross-compile-using-android-ndk
-    # llama.cpp will deal with the cpu features
-    # build_library("rnllama_v8_7" "-march=armv8.7-a")
-    # TODO: Add support runtime check for cpu features
-    # At the moment runtime check is failing.
-
-elseif (${ANDROID_ABI} STREQUAL "x86_64")
-    # x86_64 target
-    build_library("rnllama_x86_64" "-march=x86-64" "-mtune=intel" "-msse4.2" "-mpopcnt")
-endif ()
+cmake_minimum_required(VERSION 3.10)
+
+project(llama.rn)
+
+find_program(CCACHE_FOUND ccache)
+if(CCACHE_FOUND)
+  set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ccache)
+  set_property(GLOBAL PROPERTY RULE_LAUNCH_LINK ccache)
+endif(CCACHE_FOUND)
+
+set(CMAKE_CXX_STANDARD 17)
+set(RNLLAMA_LIB_DIR ${CMAKE_SOURCE_DIR}/../../../cpp)
+
+include_directories(
+    ${RNLLAMA_LIB_DIR}
+    ${RNLLAMA_LIB_DIR}/ggml-cpu
+)
+
+set(
+    SOURCE_FILES
+    ${RNLLAMA_LIB_DIR}/ggml.c
+    ${RNLLAMA_LIB_DIR}/ggml-alloc.c
+    ${RNLLAMA_LIB_DIR}/ggml-backend.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-backend-reg.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/amx/amx.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/amx/mmq.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/ggml-cpu.c
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/ggml-cpu.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/ggml-cpu-aarch64.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/ggml-cpu-quants.c
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/ggml-cpu-traits.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/unary-ops.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/binary-ops.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/sgemm.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/vec.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-cpu/ops.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-opt.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-threading.cpp
+    ${RNLLAMA_LIB_DIR}/ggml-quants.c
+    ${RNLLAMA_LIB_DIR}/gguf.cpp
+    ${RNLLAMA_LIB_DIR}/log.cpp
+    ${RNLLAMA_LIB_DIR}/llama-impl.cpp
+    ${RNLLAMA_LIB_DIR}/llama-grammar.cpp
+    ${RNLLAMA_LIB_DIR}/llama-sampling.cpp
+    ${RNLLAMA_LIB_DIR}/llama-vocab.cpp
+    ${RNLLAMA_LIB_DIR}/llama-adapter.cpp
+    ${RNLLAMA_LIB_DIR}/llama-chat.cpp
+    ${RNLLAMA_LIB_DIR}/llama-context.cpp
+    ${RNLLAMA_LIB_DIR}/llama-kv-cache.cpp
+    ${RNLLAMA_LIB_DIR}/llama-arch.cpp
+    ${RNLLAMA_LIB_DIR}/llama-batch.cpp
+    ${RNLLAMA_LIB_DIR}/llama-cparams.cpp
+    ${RNLLAMA_LIB_DIR}/llama-hparams.cpp
+    ${RNLLAMA_LIB_DIR}/llama.cpp
+    ${RNLLAMA_LIB_DIR}/llama-model.cpp
+    ${RNLLAMA_LIB_DIR}/llama-model-loader.cpp
+    ${RNLLAMA_LIB_DIR}/llama-mmap.cpp
+    ${RNLLAMA_LIB_DIR}/llama-vocab.cpp
+    ${RNLLAMA_LIB_DIR}/llama-memory.cpp
+    ${RNLLAMA_LIB_DIR}/llama-io.cpp
+    ${RNLLAMA_LIB_DIR}/llama-graph.cpp
+    ${RNLLAMA_LIB_DIR}/sampling.cpp
+    ${RNLLAMA_LIB_DIR}/unicode-data.cpp
+    ${RNLLAMA_LIB_DIR}/unicode.cpp
+    ${RNLLAMA_LIB_DIR}/common.cpp
+    ${RNLLAMA_LIB_DIR}/chat.cpp
+    ${RNLLAMA_LIB_DIR}/json-schema-to-grammar.cpp
+    ${RNLLAMA_LIB_DIR}/json.hpp
+    ${RNLLAMA_LIB_DIR}/minja/minja.hpp
+    ${RNLLAMA_LIB_DIR}/minja/chat-template.hpp
+    ${RNLLAMA_LIB_DIR}/rn-llama.cpp
+    ${CMAKE_SOURCE_DIR}/jni-utils.h
+    ${CMAKE_SOURCE_DIR}/jni.cpp
+)
+
+find_library(LOG_LIB log)
+
+function(build_library target_name cpu_flags)
+    add_library(
+        ${target_name}
+        SHARED
+        ${SOURCE_FILES}
+    )
+
+    target_link_libraries(${target_name} ${LOG_LIB} android)
+
+    target_compile_options(${target_name} PRIVATE -DLM_GGML_USE_CPU -DLM_GGML_USE_CPU_AARCH64 -pthread ${cpu_flags})
+
+    if (${CMAKE_BUILD_TYPE} STREQUAL "Debug")
+        target_compile_options(${target_name} PRIVATE -DRNLLAMA_ANDROID_ENABLE_LOGGING)
+    endif ()
+
+    # NOTE: If you want to debug the native code, you can uncomment if and endif
+    # Note that it will be extremely slow
+    # if (NOT ${CMAKE_BUILD_TYPE} STREQUAL "Debug")
+        target_compile_options(${target_name} PRIVATE -O3 -DNDEBUG -DRNLLAMA_USE_FD_FILE)
+        target_compile_options(${target_name} PRIVATE -fvisibility=hidden -fvisibility-inlines-hidden)
+        target_compile_options(${target_name} PRIVATE -ffunction-sections -fdata-sections)
+
+        target_link_options(${target_name} PRIVATE -Wl,--gc-sections)
+        target_link_options(${target_name} PRIVATE -Wl,--exclude-libs,ALL)
+        target_link_options(${target_name} PRIVATE -flto)
+    # endif ()
+endfunction()
+
+
+# Default target (no specific CPU features)
+build_library("rnllama" "")
+
+if (${ANDROID_ABI} STREQUAL "arm64-v8a")
+    # ARM64 targets
+    # Removing fp16 for now as it leads to issues with some models like deepseek r1 distills
+    # https://github.com/mybigday/llama.rn/pull/110#issuecomment-2609918310
+    build_library("rnllama_v8" "-march=armv8-a")
+    build_library("rnllama_v8_2" "-march=armv8.2-a")
+    build_library("rnllama_v8_2_dotprod" "-march=armv8.2-a+dotprod")
+    build_library("rnllama_v8_2_i8mm" "-march=armv8.2-a+i8mm")
+    build_library("rnllama_v8_2_dotprod_i8mm" "-march=armv8.2-a+dotprod+i8mm")
+
+    # https://github.com/ggerganov/llama.cpp/blob/master/docs/android.md#cross-compile-using-android-ndk
+    # llama.cpp will deal with the cpu features
+    # build_library("rnllama_v8_7" "-march=armv8.7-a")
+    # TODO: Add support runtime check for cpu features
+    # At the moment runtime check is failing.
+
+elseif (${ANDROID_ABI} STREQUAL "x86_64")
+    # x86_64 target
+    build_library("rnllama_x86_64" "-march=x86-64" "-mtune=intel" "-msse4.2" "-mpopcnt")
+endif ()