RFT_Countdown

🚀 MSBuild RFT Demo Notebook

Welcome to the Responsible Finetuning (RFT) Demo for Microsoft Build! This guide will walk you through setting up your environment and running the demo.ipynb notebook.

---

📋 Prerequisites

Before you begin, ensure you have the following installed:

• Python (3.8 or higher recommended)
• pip (Python package installer)
• Jupyter Notebook or JupyterLab

---

🛠️ Environment Setup

Follow these steps to prepare your environment:

1. Clone the Repository (if applicable) If you haven't already, clone the repository containing this demo to your local machine.

   # Example: git clone <repository-url>
   # cd <repository-folder>/MSBuildRFTDemo

2. Install Dependencies 📦 Open your terminal or command prompt in the MSBuildRFTDemo folder (where demo.ipynb and requirements.txt are located) and run:

   pip install -r requirements.txt

   This will install all the necessary Python libraries for the RFT demo.

3. Configure Environment Variables 🔑 You'll need to set up API keys and other configurations. Create a file named .env in the MSBuildRFTDemo folder and add the following content, replacing the placeholders with your actual values:

   AZURE_API_KEY="YOUR_AZURE_OPENAI_API_KEY_HERE"
   AZURE_API_ENDPOINT="YOUR_AZURE_OPENAI_ENDPOINT_HERE"
   API_VERSION="YOUR_AZURE_OPENAI_API_VERSION_HERE" # e.g., 2023-07-01-preview

   ⚠️ Important: Do not commit your .env file to version control if it contains sensitive credentials. Ensure it's listed in your .gitignore file.

---

🌟 Running the Demo Notebook

Once your environment is set up, you can run the demo:

1. Launch Jupyter Notebook 🚀 In your terminal, navigate to the MSBuildRFTDemo folder and start Jupyter Notebook:

   jupyter notebook

   Or, if you prefer JupyterLab:

   jupyter lab

   This will open Jupyter in your web browser.

2. Open demo.ipynb 📂 In the Jupyter interface, click on demo.ipynb to open the notebook.

3. Execute the Cells ▶️ Run the cells in the notebook sequentially. You can do this by:

   • Clicking the "Run" button in the toolbar for each cell.
   • Using the keyboard shortcut Shift + Enter to run the current cell and move to the next.
   • Selecting "Cell" > "Run All" from the menu bar to run all cells at once.

   The notebook will guide you through:

   • 📊 Loading and preparing data for finetuning.
   • 🛡️ Applying Responsible AI techniques during the finetuning process.
   • 🤖 Evaluating the performance and safety of the finetuned model.
   • 📈 Visualizing the results and RFT metrics.

---

📝 Notes & Troubleshooting

• API Key Errors: If you encounter errors related to API keys, double-check your .env file for typos or incorrect values. Ensure the keys have the necessary permissions for Azure OpenAI and any other services used.
• Missing Modules: If you get ModuleNotFoundError, ensure you have run pip install -r requirements.txt in the correct Python environment.
• Kernel Issues: If the notebook kernel has problems, try restarting it ("Kernel" > "Restart" in the Jupyter menu).
• RFT Specifics: This demo utilizes specific RFT tools and configurations. Refer to the comments within demo.ipynb for detailed explanations of each step.

---

Thank you for checking out the RFT Demo at Microsoft Build! ✨