ZETIC.MLange API

ZETIC.MLange is an On-device AI runtime library supporting heterogeneous Mobile NPUs’ utilization.

We provide the real On-device AI immediately with key features below

1. End-to-end on-device AI

2. One-stop deployment

3. Heterogeneous OS and target Hardware supporting

In these documentations, we provide instructions and examples for the ZETIC.MLange. Since we are keep developing very hardly with speed, Please contact ZETIC.ai for any kind of issues.

This is the `Beta` version of the ZETIC.MLange before official release.

---

Quick Start

1. Go to mlange.zetic.ai and generate Model Key and Personal Key

   • Generating Model Key Example

   # Generating Model Key with CLI Method.
   $ zetic gen -p $PROJECT_NAME -i $INPUT_0 -i $INPUT_1 .... $MODEL_PATH
   

   • Copying Personal Key Example

   

   • For more detail, refer to folowing documentations:

     • Generating Model Key

     • Generating Personal Key

2. Use your model key in Mobile Applications

• Android - Kotlin

val model = ZeticMLangeModel(this, "YOUR_PERSONAL_KEY", "YOUR_MODEL_KEY")
model.run(YOUR_INPUT_BYTE_BUFFERS)

• iOS - Swift

let model = try ZeticMLangeModel("YOUR_PERSONAL_KEY", "YOUR_MODEL_KEY")
model.run(YOUR_INPUT_DATA_ARRAY)

Contents

Overview

• What is ZETIC.MLange?
• How ZETIC.MLange Works
• Global Device Benchmark
• (Beta) ZETIC.MLange LLM Model

Getting Started

• Prepare Model and Input(s)
  • Save Model and Input(s)
  • Check the Order of Model Input(s)
• Generate Model Key
  • Terms about Generating Key
• Generate Personal Key
  • How to Generate a Personal Key Using Web Dashborad

App Implementation

• Android
• iOS

Examples

• Object Detection (YOLOv8 / YOLOv11)
• Face Detection
• Face Landmark
• Face Emotion Recognition (EMO-AffectNet)
• Audio Classification (YAMNet)
• Speech Recognition (Whisper)

Additional features

• Future Works

Index | Search Page