[go: nahoru, domu]
Skip to content

Latest commit

 

History

History

android

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
app
app
 
 
gradle/wrapper
gradle/wrapper
 
 
.gitignore
.gitignore
 
 
README.md
README.md
 
 
build.gradle.kts
build.gradle.kts
 
 
gradle.properties
gradle.properties
 
 
gradlew
gradlew
 
 
gradlew.bat
gradlew.bat
 
 
llm_inference.png
llm_inference.png
 
 
settings.gradle.kts
settings.gradle.kts
 
 

README.md

MediaPipe LLM Inference Android Demo

Overview

This is a sample app that demonstrates how to use the LLM Inference API to run common text-to-text generation tasks like information retrieval, email drafting, and document summarization.

This application must be run on a physical Android device to take advantage of the device GPU.

LLM Inference Demo

Build the demo using Android Studio

Prerequisites

  • The Android Studio IDE. This sample has been tested on Android Studio Hedgehog.

  • A physical Android device with a minimum OS version of SDK 24 (Android 7.0 - Nougat) with developer mode enabled.

Building

  • Open Android Studio. From the Welcome screen, select Open an existing Android Studio project.

  • From the Open File or Project window that appears, navigate to and select the mediapipe/examples/llm_inference/android directory. Click OK. You may be asked if you trust the project. Select Trust.

  • If it asks you to do a Gradle Sync, click OK.

  • With your Android device connected to your computer and developer mode enabled, click on the green Run arrow in Android Studio.

Models used

You can download one of the compatible models.

Once you download it, place it under the path defined as MODEL_PATH in InferenceModel on the Android device (eg. /data/local/tmp/llm/model.bin).

The easiest way to do that would be to use Android Studio's Device Explorer.

For more details, see the models section in the LLM Inference guide for Android.