Prompt Engineering Playground

This project is for experimenting with prompt engineering techniques for Large Language Models (LLMs) using the Gemini API.

Structure

• src/: Python source code
  • main.py: Main script to run experiments.
  • prompt_tech/: Core modules for the project.
    • api.py: Handles interaction with the Gemini API.
    • runner.py: Manages the execution of experiments and saving results.
• prompts/: Store prompt templates and examples in text files.
• data/: Input data for your experiments (e.g., CSV files).
• results/: Output of your experiments (e.g., JSON files with model responses).
• tests/: Tests for your experiment code.

Setup

1. Install dependencies:

   uv pip install -r requirements.txt

2. Set up your environment variables:

   • Create a .env file in the root of the project.
   • Add your Gemini API key to the .env file:

     GEMINI_API_KEY="YOUR_API_KEY"
     

Usage

1. Add your prompts:

   • You can add new prompts and techniques to the prompts dictionary in src/main.py.
   • For more complex prompts, you can save them as text files in the prompts/ directory and read them in your code.

2. Run the experiments:

   python src/main.py

3. Analyze the results:

   • The results of the experiments will be saved in results/experiment_results.json.
   • This file will contain the prompt, the model's response, the technique used, and the latency for each experiment.