Exponent-ML

Prompt + Dataset → Trained + Deployed ML Models in One Line

Exponent-ML is a CLI tool that lets anyone create, train, and deploy machine learning models by describing their task and uploading a dataset. The tool uses LLMs to generate runnable training pipelines based on both user intent and real dataset structure, with optional deployment to GitHub or cloud platforms.

🚀 Quick Start

Installation

# Clone the repository
git clone https://github.com/yourusername/exponent-ml.git
cd exponent-ml

# Install dependencies
pip install -r requirements.txt

# Set up environment variables
cp .env.example .env
# Edit .env with your API keys

Environment Setup

Create a .env file with your API keys:

# Required
ANTHROPIC_API_KEY=your_anthropic_api_key

# Authentication (OAuth)
GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_client_secret
GITHUB_CLIENT_ID=your_github_client_id
GITHUB_CLIENT_SECRET=your_github_client_secret

# Optional (for cloud features)
AWS_ACCESS_KEY_ID=your_aws_access_key
AWS_SECRET_ACCESS_KEY=your_aws_secret_key
AWS_REGION=us-east-1
S3_BUCKET=your_s3_bucket_name
MODAL_TOKEN_ID=your_modal_token_id
MODAL_TOKEN_SECRET=your_modal_token_secret
GITHUB_TOKEN=your_github_token

Note: Only ANTHROPIC_API_KEY is required. OAuth credentials are needed for authentication.

Authentication Setup

Before using Exponent-ML, you need to set up OAuth authentication:

Option 1: Use the setup script

# Set up Google OAuth
python scripts/setup_oauth.py google

# Set up GitHub OAuth  
python scripts/setup_oauth.py github

# Check your configuration
python scripts/setup_oauth.py check

Option 2: Manual setup

1. Google OAuth: Go to Google Cloud Console

   • Create a new project
   • Enable Google+ API
   • Create OAuth 2.0 credentials
   • Set redirect URI to http://localhost:8080

2. GitHub OAuth: Go to GitHub OAuth Apps

   • Create a new OAuth App
   • Set callback URL to http://localhost:8080

3. Add credentials to your .env file:

GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_client_secret
GITHUB_CLIENT_ID=your_github_client_id
GITHUB_CLIENT_SECRET=your_github_client_secret

Basic Usage

# Login with OAuth
exponent login --provider google
exponent login --provider github

# Check authentication status
exponent status

# Interactive wizard
exponent init

# Quick start with task and dataset
exponent init quick "Predict email spam" --dataset spam.csv

# Train a model
exponent train

# Deploy to GitHub
exponent deploy

📋 Commands

exponent init

Initialize a new ML project with interactive wizard:

exponent init

Options:

• --task, -t: ML task description
• --dataset, -d: Path to dataset file
• --interactive, -i: Run in interactive mode (default: True)

Subcommands:

• exponent init quick <task>: Quick initialization without prompts

exponent upload-dataset

Analyze datasets and optionally upload to S3 for cloud training:

exponent upload-dataset spam.csv

Options:

• --project-id, -p: Project ID for organization
• --upload, -u: Upload to S3 for cloud training

Subcommands:

• exponent upload-dataset analyze <file>: Analyze dataset without uploading

exponent train

Train ML models locally or in the cloud:

exponent train

Options:

• --project-id, -p: Project ID to train
• --dataset, -d: Path to dataset file
• --task, -t: Task description
• --cloud, -c: Use cloud training (Modal)
• --s3: Upload dataset to S3 for cloud training

Subcommands:

• exponent train status <job_id>: Check training job status
• exponent train list: List all training jobs

exponent deploy

Deploy projects to GitHub:

exponent deploy

Options:

• --project-id, -p: Project ID to deploy
• --name, -n: GitHub repository name
• --path: Path to project directory

Subcommands:

• exponent deploy list: List GitHub repositories

🧠 How It Works

1. Task Description: You describe your ML task in natural language
2. Dataset Analysis: The tool analyzes your dataset structure (columns, types, etc.)
3. Code Generation: An LLM generates production-ready Python code based on your task and dataset
4. Dynamic Training: The generated code is executed in Modal's cloud infrastructure
5. Deployment: Projects can be deployed to GitHub with automated workflows

No hard-coded templates - every model is generated specifically for your task and dataset!

📁 Generated Project Structure

~/.exponent/<project-id>/
├── model.py          # Model definition and training pipeline
├── train.py          # Training script with data loading
├── predict.py        # Prediction script for making predictions
├── requirements.txt  # Python dependencies
└── README.md        # Project documentation

🔧 Configuration

Required Environment Variables

• ANTHROPIC_API_KEY: Your Anthropic API key for code generation

Optional Environment Variables

• AWS_ACCESS_KEY_ID: AWS access key for S3 uploads
• AWS_SECRET_ACCESS_KEY: AWS secret key for S3 uploads
• AWS_REGION: AWS region (default: us-east-1)
• S3_BUCKET: S3 bucket name for dataset storage
• MODAL_TOKEN_ID: Modal token for cloud training
• MODAL_TOKEN_SECRET: Modal token secret for cloud training
• GITHUB_TOKEN: GitHub token for repository creation

🛠️ Development

Project Structure

exponent-ml/
├── exponent/
│   ├── cli/              # CLI interface (Typer)
│   │   └── commands/     # CLI commands
│   ├── core/             # Core logic
│   │   ├── code_gen.py   # LLM code generation
│   │   ├── config.py     # Configuration management
│   │   ├── s3_utils.py   # S3 dataset handling
│   │   ├── modal_runner.py # Modal cloud training
│   │   └── github_utils.py # GitHub deployment
│   └── main.py           # CLI entry point
├── requirements.txt       # Python dependencies
├── pyproject.toml        # Project configuration
└── README.md            # This file

Local Development

# Install in development mode
pip install -e .

# Run tests
pytest

# Format code
black exponent/
isort exponent/

# Lint code
flake8 exponent/

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request

📄 License

MIT License - see LICENSE file for details.

🙏 Acknowledgments

• Anthropic for Claude API
• Modal for cloud infrastructure
• Typer for CLI framework
• scikit-learn for ML algorithms