Command Line Interface

This guide explains how to deploy your models via terminal using the ZETIC CLI.

You must first create a repository using the Web Dashboard before using the CLI.

Step 1: Install ZETIC CLI

Install the ZETIC CLI using pip:

pip install zetic

Verify the installation:

zetic --help

Step 2: Authenticate

zetic auth login

This will open a browser window for authentication. Once completed, your credentials will be stored locally.

You only need to log in once. Your session will persist until you log out.

Step 3: Upload Your Model

Upload your model to an existing repository:

zetic gen -p $REPOSITORY -i $INPUT_0 -i $INPUT_1 ... $MODEL_PATH

Parameters

Parameter	Description	Example
`-p, --project`	Repository identifier in `username/repository_name` format	`steve/mobileclip-image`
`-i, --input`	Input file path(s). Supports `.npy` format only. Repeat for multiple inputs	`-i input0.npy -i input1.npy`
`MODEL_PATH`	Path to your model file. Supports `.pt2`, `.onnx`, and `.pt` (deprecated)	`model.onnx`

Ensure your inputs are in the correct order. Learn more in Supported Formats.

Examples

zetic gen \
  -p steve/mobileclip-image \
  -i input_image.npy \
  mobileclip.onnx

zetic gen \
  -p steve/llama-chat \
  -i input_ids.npy \
  -i attention_mask.npy \
  llama_model.pt2

zetic gen \
  -p steve/llama-chat \
  -i input_ids.npy \
  -i attention_mask.npy \
  llama_model.pt

Expected Output

Uploading model from MODEL_NAME.onnx to project USER_NAME/MODEL_NAME
Starting upload process...
Project: USER_NAME/MODEL_NAME
Model Path: MODEL_NAME.onnx
Upload completed successfully!
Model is requested to convert
Your model key is MODEL_KEY_HASH
Check your dashboard at: https://melange.zetic.ai/p/USER_NAME/MODEL_NAME/models/MODEL_KEY_HASH

For CI/CD pipelines, consider using Personal Keys as environment variables instead of interactive login. See CI/CD Integration for details.

Next Steps

Understanding Model Keys: Learn about model identifiers and versioning
Web Dashboard: Manage models visually
CI/CD Integration: Automate deployments