TaskEval

Setup

Prerequisites

1. Python 3.10+ - Make sure you have Python 3.10 or higher installed
2. Poetry - For dependency management
3. direnv - For environment variable management

Installation

1. Install direnv (if not already installed):

   # On macOS with Homebrew
   brew install direnv
   
   # On Ubuntu/Debian
   sudo apt-get install direnv
   
   # On other systems, check: https://direnv.net/docs/installation.html

2. Install Python dependencies:

   poetry install

3. Set up environment variables:

   • Edit .envrc and replace the placeholder API key with your actual Anthropic API key:

   export ANTHROPIC_API_KEY="your-actual-api-key-here"
   export OPENAI_API_KEY="your-actual-api-key-here"

4. Allow direnv to load the environment:

   direnv allow

Usage

Running the Application

Start the Streamlit App

To run the TaskEval Streamlit application:

# Activate the Poetry environment and run Streamlit
poetry run streamlit run app.py

Or if you prefer to activate the virtual environment first:

# Activate the Poetry shell
poetry shell

# Run the Streamlit app
streamlit run app.py

The application will start and be available at http://localhost:8501 in your web browser.

Application Features

The TaskEval app provides two main evaluation tasks:

1. Figure Extraction: Upload an image and provide instructions to extract data points from charts/graphs
2. QA Answering: Upload a PDF document and ask questions about its content

Troubleshooting

If you encounter any issues:

1. Port already in use: If port 8501 is busy, Streamlit will automatically use the next available port (8502, 8503, etc.)

2. Environment variables not loaded: Make sure you've run direnv allow in the project directory

3. Dependencies not found: Ensure you've run poetry install and activated the environment with poetry shell

4. API key issues: Verify your API keys are correctly set in the .envrc file