Zero-Data AI Model Foundry

This repository contains a Covalent Cloud Jupyter notebook that creates an AI model foundry for zero-data fine-tuning.

Context

The example here was prepared for PyCon 2024 in Pittsburgh, PA.

Instructions

To follow along

Open the Jupyter notebook in your browser (or in an IDE).

jupyter notebook zero-data-model-foundry.ipynb

To run the notebook

Running the notebook requires a Covalent Cloud account.

1. Log in and copy your API key from the Covalent Cloud dashboard. Paste the key as the value CC_API_KEY in the environment.yml file.

2. Create a conda environment using the environment.yml file.

conda env create -f environment.yml

3. Active the environment.

conda activate covalent-pycon-2024

4. Open the Jupyter notebook in your browser (or in an IDE).

jupyter notebook zero-data-model-foundry.ipynb