Add Llama server (OpenAI compatible)

app/build.gradle.kts

@@ -64,6 +64,10 @@ android {
             exclude(group = "org.jetbrains", module = "annotations")
         }
     }
+    packaging {
+        resources.excludes.add("META-INF/INDEX.LIST")
+        resources.excludes.add("META-INF/io.netty.versions.properties")
+    }
 }
 
 ksp {
@@ -84,6 +88,7 @@ dependencies {
 
     implementation(project(":smollm"))
     implementation(project(":hf-model-hub-api"))
+    implementation(project(":smollm-server"))
 
     // Koin: dependency injection
     implementation(libs.koin.android)

app/src/main/AndroidManifest.xml

@@ -43,6 +43,10 @@
         <activity android:name=".ui.screens.model_download.DownloadModelActivity"/>
 
         <activity android:name=".ui.screens.manage_tasks.ManageTasksActivity"/>
+        <service
+            android:name=".server.LlamaServerService"
+            android:exported="false"
+            />
     </application>
 
 </manifest>

app/src/main/java/io/shubham0204/smollmandroid/server/LlamaServerService.kt

@@ -0,0 +1,28 @@
+package io.shubham0204.smollmandroid.server
+
+import android.app.Service
+import android.content.Intent
+import android.os.IBinder
+import io.shubham0204.smollm_server.LlamaServer
+
+class LlamaServerService : Service() {
+    private val llamaServer = LlamaServer()
+
+    override fun onBind(intent: Intent?): IBinder? {
+        TODO("Not yet implemented")
+    }
+
+    override fun onStartCommand(
+        intent: Intent?,
+        flags: Int,
+        startId: Int,
+    ): Int {
+        llamaServer.start()
+        return START_STICKY
+    }
+
+    override fun onDestroy() {
+        llamaServer.stop()
+        super.onDestroy()
+    }
+}

app/src/main/java/io/shubham0204/smollmandroid/ui/screens/chat/ChatActivity.kt

@@ -100,6 +100,7 @@ import androidx.navigation.compose.rememberNavController
 import io.shubham0204.smollmandroid.R
 import io.shubham0204.smollmandroid.data.Chat
 import io.shubham0204.smollmandroid.data.Task
+import io.shubham0204.smollmandroid.server.LlamaServerService
 import io.shubham0204.smollmandroid.ui.components.AppBarTitleText
 import io.shubham0204.smollmandroid.ui.components.MediumLabelText
 import io.shubham0204.smollmandroid.ui.screens.manage_tasks.ManageTasksActivity
@@ -143,6 +144,9 @@ class ChatActivity : ComponentActivity() {
             }
         }
 
+
+        startService(Intent(this, LlamaServerService::class.java))
+
         setContent {
             val navController = rememberNavController()
             NavHost(

build.gradle.kts

@@ -4,7 +4,7 @@ plugins {
     alias(libs.plugins.kotlin.android) apply false
     alias(libs.plugins.kotlin.compose) apply false
     alias(libs.plugins.android.library) apply false
-    id("com.google.devtools.ksp") version "2.0.0-1.0.24" apply false
+    id("com.google.devtools.ksp") version "2.0.21-1.0.28" apply false
     alias(libs.plugins.jetbrains.kotlin.jvm) apply false
-    kotlin("plugin.serialization") version "2.1.0" apply false
+    kotlin("plugin.serialization") version "2.1.10" apply false
 }

gradle/libs.versions.toml

@@ -13,6 +13,8 @@ koin = "3.5.6"
 koinAnnotations = "1.3.1"
 jetbrainsKotlinJvm = "2.0.0"
 uiTextGoogleFonts = "1.7.7"
+appcompat = "1.7.0"
+material = "1.12.0"
 
 
 [libraries]
@@ -39,6 +41,8 @@ koin-androidx-compose = { module = "io.insert-koin:koin-androidx-compose", versi
 koin-annotations = { module = "io.insert-koin:koin-annotations", version.ref = "koinAnnotations" }
 koin-ksp-compiler = { module = "io.insert-koin:koin-ksp-compiler", version.ref = "koinAnnotations" }
 androidx-ui-text-google-fonts = { group = "androidx.compose.ui", name = "ui-text-google-fonts", version.ref = "uiTextGoogleFonts" }
+androidx-appcompat = { group = "androidx.appcompat", name = "appcompat", version.ref = "appcompat" }
+material = { group = "com.google.android.material", name = "material", version.ref = "material" }
 
 
 [plugins]

hf-model-hub-api/build.gradle.kts

@@ -1,7 +1,7 @@
 plugins {
     id("java-library")
     alias(libs.plugins.jetbrains.kotlin.jvm)
-    kotlin("plugin.serialization") version "2.1.0"
+    kotlin("plugin.serialization") version "2.1.10"
 }
 
 val ktorVersion = "3.0.2"

settings.gradle.kts

@@ -1,5 +1,8 @@
 import java.net.URI
 
+include(":smollm-server")
+
+
 include(":hf-model-hub-api")
 
 

smollm-server/.gitignore

@@ -0,0 +1 @@
+/build

smollm-server/build.gradle.kts

@@ -0,0 +1,59 @@
+plugins {
+    alias(libs.plugins.android.library)
+    alias(libs.plugins.kotlin.android)
+    kotlin("plugin.serialization") version "2.1.10"
+    id("com.google.devtools.ksp")
+}
+
+android {
+    namespace = "io.shubham0204.smollm_server"
+    compileSdk = 35
+
+    defaultConfig {
+        minSdk = 24
+
+        testInstrumentationRunner = "androidx.test.runner.AndroidJUnitRunner"
+        consumerProguardFiles("consumer-rules.pro")
+    }
+
+    buildTypes {
+        release {
+            isMinifyEnabled = false
+            proguardFiles(
+                getDefaultProguardFile("proguard-android-optimize.txt"),
+                "proguard-rules.pro",
+            )
+        }
+    }
+    compileOptions {
+        sourceCompatibility = JavaVersion.VERSION_21
+        targetCompatibility = JavaVersion.VERSION_21
+    }
+    kotlinOptions {
+        jvmTarget = "21"
+    }
+}
+
+val ktorVersion = "3.0.3"
+
+dependencies {
+    implementation(libs.androidx.core.ktx)
+    implementation(libs.androidx.appcompat)
+    implementation(libs.material)
+
+    // Ktor server
+    implementation("io.ktor:ktor-server-core:$ktorVersion")
+    implementation("io.ktor:ktor-server-netty:$ktorVersion")
+
+    // kotlinx serialization (JSON) and
+    // content-negotiation
+    // docs: https://ktor.io/docs/server-serialization.html
+    implementation("io.ktor:ktor-server-content-negotiation:$ktorVersion")
+    implementation("io.ktor:ktor-serialization-kotlinx-json:$ktorVersion")
+
+    implementation(project(":smollm"))
+
+    testImplementation(libs.junit)
+    androidTestImplementation(libs.androidx.junit)
+    androidTestImplementation(libs.androidx.espresso.core)
+}

smollm-server/proguard-rules.pro

@@ -0,0 +1,21 @@
+# Add project specific ProGuard rules here.
+# You can control the set of applied configuration files using the
+# proguardFiles setting in build.gradle.
+#
+# For more details, see
+#   http://developer.android.com/guide/developing/tools/proguard.html
+
+# If your project uses WebView with JS, uncomment the following
+# and specify the fully qualified class name to the JavaScript interface
+# class:
+#-keepclassmembers class fqcn.of.javascript.interface.for.webview {
+#   public *;
+#}
+
+# Uncomment this to preserve the line number information for
+# debugging stack traces.
+#-keepattributes SourceFile,LineNumberTable
+
+# If you keep the line number information, uncomment this to
+# hide the original source file name.
+#-renamesourcefileattribute SourceFile

smollm-server/src/androidTest/java/io/shubham0204/smollm_server/ExampleInstrumentedTest.kt

@@ -0,0 +1,24 @@
+package io.shubham0204.smollm_server
+
+import androidx.test.platform.app.InstrumentationRegistry
+import androidx.test.ext.junit.runners.AndroidJUnit4
+
+import org.junit.Test
+import org.junit.runner.RunWith
+
+import org.junit.Assert.*
+
+/**
+ * Instrumented test, which will execute on an Android device.
+ *
+ * See [testing documentation](http://d.android.com/tools/testing).
+ */
+@RunWith(AndroidJUnit4::class)
+class ExampleInstrumentedTest {
+    @Test
+    fun useAppContext() {
+        // Context of the app under test.
+        val appContext = InstrumentationRegistry.getInstrumentation().targetContext
+        assertEquals("io.shubham0204.smollm_server.test", appContext.packageName)
+    }
+}

smollm-server/src/main/AndroidManifest.xml

@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="utf-8"?>
+<manifest xmlns:android="http://schemas.android.com/apk/res/android">
+
+</manifest>

smollm-server/src/main/java/io/shubham0204/smollm_server/LlamaServer.kt

@@ -0,0 +1,78 @@
+package io.shubham0204.smollm_server
+
+import io.ktor.http.HttpStatusCode
+import io.ktor.serialization.kotlinx.json.json
+import io.ktor.server.application.Application
+import io.ktor.server.application.install
+import io.ktor.server.engine.connector
+import io.ktor.server.engine.embeddedServer
+import io.ktor.server.netty.Netty
+import io.ktor.server.plugins.contentnegotiation.ContentNegotiation
+import io.ktor.server.request.receive
+import io.ktor.server.response.respond
+import io.ktor.server.routing.get
+import io.ktor.server.routing.post
+import io.ktor.server.routing.route
+import io.ktor.server.routing.routing
+import io.shubham0204.smollm_server.models.OpenAIChatCompletionRequest
+import kotlinx.serialization.json.Json
+
+class LlamaServer {
+    private val server =
+        embeddedServer(
+            factory = Netty,
+            configure = {
+                connector {
+                    port = 8080
+                }
+            },
+        ) {
+            setupPlugins()
+            setupRoutes()
+        }
+
+    fun start() {
+        server.start(wait = false)
+    }
+
+    fun stop() {
+        server.stop()
+    }
+
+    private fun Application.setupPlugins() {
+        // install content-negotiation and
+        // configure JSON serializer
+        // docs: https://ktor.io/docs/server-serialization.html#install_plugin
+        //       https://ktor.io/docs/server-serialization.html#register_xml
+        install(ContentNegotiation) {
+            json(
+                Json {
+                    isLenient = true
+                    prettyPrint = true
+                },
+            )
+        }
+    }
+
+    private fun Application.setupRoutes() {
+        routing {
+            get {
+                call.respond(HttpStatusCode.OK, "Hello, world!")
+            }
+            route("v1") {
+                route("chat") {
+                    post("completions") {
+                        val request = call.receive<OpenAIChatCompletionRequest>()
+                    }
+                }
+                get("/echo") {
+                    val message =
+                        call.queryParameters["message"] ?: call.respond(HttpStatusCode.BadRequest)
+                    call.respond(HttpStatusCode.OK, message)
+                }
+                get("/models") {
+                }
+            }
+        }
+    }
+}

smollm-server/src/main/java/io/shubham0204/smollm_server/models/OpenAIChatCompletionRequest.kt

@@ -0,0 +1,25 @@
+package io.shubham0204.smollm_server.models
+
+import kotlinx.serialization.Serializable
+
+/**
+ * Represents a message in the OpenAI Chat Completion API
+ * docs: https://platform.openai.com/docs/api-reference/chat/create#chat-create-messages
+ */
+@Serializable
+data class ChatCompletionMessage(
+    val role: String,
+    val content: String,
+)
+
+/**
+ * Represents the request for the OpenAI Chat Completion API
+ * docs: https://platform.openai.com/docs/api-reference/chat/create
+ */
+@Serializable
+data class OpenAIChatCompletionRequest(
+    val messages: List<ChatCompletionMessage>,
+    val model: String,
+    val topP: Float = 1.0f,
+    val temperature: Float = 1.0f,
+)

smollm-server/src/main/java/io/shubham0204/smollm_server/models/OpenAIModel.kt

@@ -0,0 +1,16 @@
+package io.shubham0204.smollm_server.models
+
+import kotlinx.serialization.SerialName
+import kotlinx.serialization.Serializable
+
+/**
+ * OpenAI Model object, representing a LLM model available for inference
+ * docs: https://platform.openai.com/docs/api-reference/models/object
+ */
+@Serializable
+data class OpenAIModel(
+    val id: String,
+    val created: Long,
+    @SerialName("object") val object_: String = "model", // `object` is a reserved word in Kotlin
+    @SerialName("owned_by") val ownedBy: String,
+)

smollm-server/src/test/java/io/shubham0204/smollm_server/ExampleUnitTest.kt

@@ -0,0 +1,17 @@
+package io.shubham0204.smollm_server
+
+import org.junit.Test
+
+import org.junit.Assert.*
+
+/**
+ * Example local unit test, which will execute on the development machine (host).
+ *
+ * See [testing documentation](http://d.android.com/tools/testing).
+ */
+class ExampleUnitTest {
+    @Test
+    fun addition_isCorrect() {
+        assertEquals(4, 2 + 2)
+    }
+}