Quick Start: Run a Model Zoo Model Locally

This guide walks you through setting up the Bitweaver SDK locally, then validating a Model Zoo model on your machine. Once you have a validated model, continue to Your First Benchmark to upload it to Bitweaver and run it on real hardware.

Don't have a Bitweaver account yet? Get in touch.

1. Install the dg SDK

pip install deepgate

The package is deepgate on PyPI; the import name is dg.

2. Copy the Validate script from the Model Zoo

In Bitweaver, navigate to Model Zoo → choose a model → Validate. Copy the generated script into your IDE.

The script includes all the imports, validations, and schema save logic you need.

3. Run the script and review results

Run the script in your Python environment and inspect the output.

4. Save the schema JSON

The script writes a schema JSON to your working directory. Keep this file - it is what Bitweaver uses to compile and benchmark the model.

5. Watch it end-to-end

Next steps

You now have a validated model and its schema JSON. Continue to:

• Your First Benchmark - upload the model, pick target boards, and review on-MCU performance.
• End-to-End Tutorial - train a model with the dg SDK and deploy it through Bitweaver.