Merge branch 'main' into copilot/add-instrumentation-test-workflow

Author: kirklandsign

.github/workflows/android-build.yml

@@ -10,7 +10,15 @@ on:
   push:
   pull_request:
     branches: [main]
+  schedule:
+    # Run nightly at midnight UTC
+    - cron: '0 0 * * *'
   workflow_dispatch:
+    inputs:
+      local_aar:
+        description: 'URL to download a local AAR file. When set, the workflow will download the AAR and use it instead of the Maven dependency.'
+        required: false
+        type: string
 
 permissions:
   contents: read
@@ -41,9 +49,20 @@ jobs:
       - name: Setup Gradle
         uses: gradle/actions/setup-gradle@v4
 
+      - name: Download local AAR
+        if: ${{ inputs.local_aar && matrix.name == 'LlamaDemo' }}
+        run: |
+          mkdir -p ${{ matrix.path }}/app/libs
+          curl -fL -o ${{ matrix.path }}/app/libs/executorch.aar "${{ inputs.local_aar }}"
+
       - name: Build with Gradle
         working-directory: ${{ matrix.path }}
-        run: ./gradlew build --no-daemon
+        run: |
+          if [ -n "${{ inputs.local_aar }}" ] && [ "${{ matrix.name }}" == "LlamaDemo" ]; then
+            ./gradlew build --no-daemon -PuseLocalAar=true
+          else
+            ./gradlew build --no-daemon
+          fi
 
       - name: Upload build artifacts
         uses: actions/upload-artifact@v4

.github/workflows/export-models.yml

@@ -0,0 +1,67 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+name: Export Models
+
+on:
+  schedule:
+    # Run nightly at midnight UTC
+    - cron: '0 0 * * *'
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  export:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - name: dl3
+            path: dl3/python
+            artifact: dl3-xnnpack-fp32
+            output: dl3/python/dl3_xnnpack_fp32.pte
+          - name: mv2
+            path: mv2/python
+            artifact: mv2-xnnpack
+            output: mv2/python/model_mv2_xnnpack.pte
+          - name: mv3
+            path: mv3/python
+            artifact: mv3-models
+            output: |
+              mv3/python/mv3.pte
+              mv3/python/mv3_coreml_all.pte
+              mv3/python/mv3_mps_float16.pte
+              mv3/python/mv3_xnnpack_fp32.pte
+
+    name: Export ${{ matrix.name }}
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install executorch torchvision
+
+      - name: Run export script
+        working-directory: ${{ matrix.path }}
+        run: python export.py
+
+      - name: Upload exported model
+        uses: actions/upload-artifact@v4
+        with:
+          name: ${{ matrix.artifact }}
+          path: ${{ matrix.output }}
+          if-no-files-found: error

cifar/android/CifarETTrainingDemo/gradle/wrapper/gradle-wrapper.jar

@@ -15,16 +15,17 @@ The goal is for you to see the type of support ExecuTorch provides and feel comf
 
 ## Supporting Models
 As a whole, the models that this app supports are (varies by delegate):
-* Llama 3.2 Quantized 1B/3B
-* Llama 3.2 1B/3B in BF16
-* Llama Guard 3 1B
-* Llama 3.1 8B
-* Llama 3 8B
-* Llama 2 7B
-* LLaVA-1.5 vision model (only XNNPACK)
-* Qwen 3 0.6B, 1.7B, and 4B
-* Voxtral Mini 3B
-* Gemma 3 4B
+* [Llama](https://github.com/pytorch/executorch/tree/main/examples/models/llama)
+      * Llama 3.2 Quantized 1B/3B
+      * Llama 3.2 1B/3B in BF16
+      * Llama Guard 3 1B
+      * Llama 3.1 8B
+      * Llama 3 8B
+      * Llama 2 7B
+* [LLaVA-1.5 vision model (only XNNPACK)](https://github.com/pytorch/executorch/tree/main/examples/models/llava)
+* [Qwen 3 0.6B, 1.7B, and 4B](https://github.com/pytorch/executorch/tree/main/examples/models/qwen3)
+* [Voxtral Mini 3B](https://github.com/pytorch/executorch/tree/main/examples/models/voxtral)
+* [Gemma 3 4B](https://github.com/pytorch/executorch/tree/main/examples/models/gemma3)
 
 ## Building the APK
 First it’s important to note that by default, the app depends on [ExecuTorch library](https://central.sonatype.com/artifact/org.pytorch/executorch-android) on Maven Central. It uses the latest `org.pytorch:executorch-android` package, which comes with all the default kernel libraries (portable, quantized, optimized), LLM customized libraries, and XNNPACK backend.
@@ -64,6 +65,18 @@ Select the settings widget to get started with picking a model, its parameters a
 
 
 
+### Push Model and Tokenizer Files to Device
+
+Before selecting a model and tokenizer in the app, you need to push these files to your Android device. Use the following commands to copy the model (`.pte`) and tokenizer files to the device:
+
+```sh
+adb shell mkdir -p /data/local/tmp/llama
+adb push <your_model>.pte /data/local/tmp/llama
+adb push <your_tokenizer> /data/local/tmp/llama
+```
+
+Replace `<your_model>.pte` with your exported model file and `<your_tokenizer>` with your tokenizer file (e.g., `tokenizer.bin` or `tokenizer.model`).
+
 ### Select Models and Parameters
 
 Once you've selected the model, tokenizer, and model type you are ready to click on "Load Model" to have the app load the model and go back to the main Chat activity.
@@ -91,15 +104,15 @@ mModule = new LlmModule(
 int loadResult = mModule.load();
 ```
 
-* `modelCategory`: Indicate whether it’s a text-only or vision model
-* `modePath`: path to the .pte file
-* `tokenizerPath`: path to the tokenizer file
-* `temperature`: model parameter to adjust the randomness of the model’s output
-* `dataPath`: path to one or a list of .ptd files
+* `modelCategory`: Indicates whether it’s a text-only or vision model
+* `modelPath`: Path to the .pte file
+* `tokenizerPath`: Path to the tokenizer file
+* `temperature`: Model parameter to adjust the randomness of the model’s output
+* `dataPath`: Path to one or a list of .ptd files
 
 
 ### User Prompt
-Once model is successfully loaded then enter any prompt and click the send (i.e. generate) button to send it to the model.
+Once the model is successfully loaded, enter any prompt and click the send (i.e., generate) button to send it to the model.
 <p align="center">
 <img src="https://raw.githubusercontent.com/pytorch/executorch/refs/heads/main/docs/source/_static/img/load_complete_and_start_prompt.png" style="width:300px">
 </p>
@@ -112,11 +125,11 @@ You can provide it more follow-up questions as well.
 #### ExecuTorch App API
 
 ```java
-mModule.generate(prompt,sequence_length, MainActivity.this);
+mModule.generate(prompt, sequence_length, MainActivity.this);
 ```
-* `prompt`: User formatted prompt
+* `prompt`: User-formatted prompt
 * `sequence_length`: Number of tokens to generate in response to a prompt
-* `MainActivity.this`: Indicate that the callback functions (OnResult(), OnStats()) are present in this class.
+* `MainActivity.this`: Indicates that the callback functions (`onResult()`, `onStats()`) are present in this class.
 
 [*LLaVA-1.5: Only for XNNPACK delegate*]
 
@@ -177,7 +190,7 @@ Ensure you have the following functions in your callback class that you provided
 ```
 
 ## Instrumentation Test
-You can run the instrumentation test for sanity check. The test loads a model pte file and tokenizer.bin file
+You can run the instrumentation test for a sanity check. The test loads a model .pte file and tokenizer.bin file
 under `/data/local/tmp/llama`.
 
 ### Model preparation
@@ -204,4 +217,4 @@ adb push tokenizer.bin /data/local/tmp/llama
 ```
 
 ## Reporting Issues
-If you encountered any bugs or issues following this tutorial please file a bug/issue here on [Github](https://github.com/pytorch/executorch/issues/new), or join our discord [here](https://lnkd.in/gWCM4ViK).
+If you encountered any bugs or issues following this tutorial, please file a bug/issue here on [GitHub](https://github.com/pytorch/executorch/issues/new), or join our Discord [here](https://lnkd.in/gWCM4ViK).

llm/android/LlamaDemo/README.md

@@ -1,12 +1,12 @@
-# Guide to set up Java/SDK/NDK for Android
+# Guide to Set Up Java/SDK/NDK for Android
 
-Follow this doc if you haven't set up Java/SDK/NDK for Android development
+Follow this guide if you haven't set up Java/SDK/NDK for Android development
 already.
-This doc provides a CLI tutorial to set them up. Otherwise, you can do the same
-thing with Android Studio GUI.
+This guide provides a CLI tutorial to set them up. Alternatively, you can do the same
+thing with the Android Studio GUI.
 
-## Set up Java 17
-1. Download the archive from Oracle website.
+## Set Up Java 17
+1. Download the archive from the Oracle website.
 Make sure you have read and agree with the terms and conditions from the website before downloading.
 ```bash
 export DEV_HOME=<path-to-dev>
@@ -40,14 +40,14 @@ export PATH="$JAVA_HOME/bin:$PATH"
 Note: Oracle has tutorials for installing Java on
 [Linux](https://docs.oracle.com/en/java/javase/17/install/installation-jdk-linux-platforms.html#GUID-4A6BD592-1840-4BB4-A758-4CD49E9EE88B)
 and [macOS](https://docs.oracle.com/en/java/javase/17/install/installation-jdk-macos.html#GUID-E8A251B6-D9A9-4276-ABC8-CC0DAD62EA33).
-Some Linux distributions has JDK package in package manager. For example, Debian users can install
-openjdk-17-jdk package.
+Some Linux distributions have a JDK package in the package manager. For example, Debian users can install
+the openjdk-17-jdk package.
 
-## Set up Android SDK/NDK
-Android has a command line tool [sdkmanager](https://developer.android.com/tools/sdkmanager) which
-helps users managing SDK and other tools related to Android development.
+## Set Up Android SDK/NDK
+Android has a command-line tool [sdkmanager](https://developer.android.com/tools/sdkmanager) which
+helps users manage SDK and other tools related to Android development.
 
-1. Go to https://developer.android.com/studio and download the archive from "Command line tools
+1. Go to https://developer.android.com/studio and download the archive from the "Command line tools
 only" section. Make sure you have read and agree with the terms and conditions from the website.
 
 Linux:
@@ -58,7 +58,7 @@ macOS:
 ```bash
 curl https://dl.google.com/android/repository/commandlinetools-mac-11076708_latest.zip -o commandlinetools.zip
 ```
-2. Unzip.
+2. Unzip:
 ```bash
 unzip commandlinetools.zip
 ```
@@ -80,7 +80,7 @@ If you want to use Android Studio and never set up Java/SDK/NDK before, or if
 you use the newly installed ones, follow these steps to set Android Studio to use
 them.
 
-Copy these output paths to be used by Android Studio
+Copy these output paths to be used by Android Studio:
 ```bash
 echo $ANDROID_HOME
 echo $ANDROID_NDK

llm/android/LlamaDemo/SDK-quick-setup-guide.md

@@ -381,11 +381,20 @@ protected void onResume() {
           // If users change the model file, but not pressing loadModelButton, we won't load the new
           // model
           checkForUpdateAndReloadModel(updatedSettingsFields);
-        } else {
+        } else if (mModule == null) {
+          // Only ask user to select model if no model is currently loaded
+          askUserToSelectModel();
+        }
+      } else {
+        // Settings not updated, but still check if model/tokenizer is not selected
+        String modelPath = updatedSettingsFields.getModelFilePath();
+        String tokenizerPath = updatedSettingsFields.getTokenizerFilePath();
+        if (modelPath.isEmpty() || tokenizerPath.isEmpty()) {
           askUserToSelectModel();
         }
       }
-    } else {
+    } else if (mModule == null) {
+      // Only ask user to select model if no model is currently loaded
       askUserToSelectModel();
     }
   }
@@ -452,8 +461,16 @@ private void askUserToSelectModel() {
     ETLogging.getInstance().log(askLoadModel);
     runOnUiThread(
         () -> {
-          mMessageAdapter.add(askLoadModelMessage);
-          mMessageAdapter.notifyDataSetChanged();
+          if (!mMessageAdapter.isDuplicateSystemMessage(askLoadModel)) {
+            mMessageAdapter.add(askLoadModelMessage);
+            mMessageAdapter.notifyDataSetChanged();
+          }
+          new AlertDialog.Builder(this)
+              .setTitle("Please Select a Model")
+              .setMessage(
+                  "Please select a model and tokenizer from the settings (top right corner) to get started.")
+              .setPositiveButton(android.R.string.ok, null)
+              .show();
         });
   }
 

llm/android/LlamaDemo/app/src/main/java/com/example/executorchllamademo/MainActivity.java

@@ -132,4 +132,21 @@ public int getMaxPromptID() {
     }
     return maxPromptID;
   }
+
+  /**
+   * Checks if the last message is a duplicate system message with the given text.
+   *
+   * @param text The text to check against the last message
+   * @return true if the last message is a system message with matching text, false otherwise
+   */
+  boolean isDuplicateSystemMessage(String text) {
+    int count = getCount();
+    if (count == 0) {
+      return false;
+    }
+    Message lastMessage = getItem(count - 1);
+    return lastMessage != null
+        && lastMessage.getMessageType() == MessageType.SYSTEM
+        && text.equals(lastMessage.getText());
+  }
 }