Deploy Your Model
Master the Melange deployment pipeline, from artifact management to on-device integration.
Deployment Pipeline Overview
Melange offers flexible interfaces for managing your MLOps workflow. Select the tool that fits your stage of development:
Web Dashboard
Interactive GUI for initial project setup and visual model management.
CLI
Headless automation tool designed for CI/CD integration.
Prerequisites
A one-time initial setup via the Web Dashboard is required before enabling CLI automation.
Artifact Management
Repository Structure
A Repository acts as the version control container for a specific model lineage. It maintains distinct versions of your artifacts.
username/my-model-repository
├── v1 (uploaded 2024-01-15)
├── v2 (uploaded 2024-02-20) ← default
└── v3 (uploaded 2024-03-10) ← latestVersioning Policy:
- Implicit Latest: Clients automatically pull the most recent upload unless pinned.
- Default Pinning: Administrators can designate a specific stable version as
defaultfor production channels. - Immutable History: All uploaded versions are preserved for rollback capabilities.
Model Identifier
Reference models programmatically using the key format: username/repository_name.
// NPU-Accelerated Model Loader
val model = ZeticMLangeModel(
context,
BuildConfig.PERSONAL_KEY,
"USERID/REPOSITORY_NAME" // Automatically resolves to 'default' or 'latest'
)// NPU-Accelerated Model Loader
let model = try ZeticMLangeModel(
personalKey: personalKey,
name: "USERID/REPOSITORY_NAME" // Automatically resolves to 'default' or 'latest'
)// NPU-Accelerated Model Loader
const model = await ZeticMLange.load(
personalKey,
"USERID/REPOSITORY_NAME"
);Deployment Status at Melange Dashboard
Monitor the compilation and optimization status of your artifacts:
| State | Phase Description | Runtime Viability |
|---|---|---|
| N/A | Repository initialized; awaiting binary upload. | ❌ |
| Failed | Validation or compilation error (e.g., ONNX Opset mismatch). | ❌ |
| Converting | Active graph lowering and quantization in progress. | ❌ |
| Optimizing | Functional binary available; throughput tuning active. | ✅ Sub-optimal performance |
| Ready | Fully compiled, tuned, and validated for production. | ✅ Production |
Optimization Latency
Models in the Optimizing state are executable but may not yet utilize the full NPU throughput. Stick to Ready artifacts for performance benchmarking.
Security & Authentication
Personal Key Management
The Personal Key is your persistent credential for SDK authentication and API access. Treat it as a high-value secret.
Provision Credential
- Access Settings → Personal Key in the Dashboard.
- Select Generate New Key.
- Copy Immediately: The key is displayed only once.
Save the key securely
Store this string in a password manager immediately. It cannot be retrieved later.
SDK Initialization
Instantiate the model.
val model = ZeticMLangeModel(
context,
"PERSONAL_KEY",
"USERID/REPOSITORY_NAME"
)let model = try ZeticMLangeModel(
personalKey: "personalKey",
name: "USERID/REPOSITORY_NAME"
)