Installation Guide

This guide will help you install and set up the JEPA framework.

Requirements

System Requirements

• Python 3.8 or higher

• CUDA 11.0+ (for GPU training)

• 8GB+ RAM (16GB+ recommended)

• 10GB+ disk space

Core Dependencies

• PyTorch 2.0+

• NumPy

• PyYAML

• tqdm

Optional Dependencies

• Weights & Biases (for experiment tracking)

• TensorBoard (for visualization)

• Triton (for kernel optimization)

• Hugging Face Transformers (for compatibility)

Installation Methods

From Source (Recommended)

1. Clone the repository:

git clone https://github.com/your-org/jepa.git
cd jepa

2. Create a virtual environment:

python -m venv jepa-env
source jepa-env/bin/activate  # On Windows: jepa-env\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Install in development mode:

pip install -e .

Using pip

pip install jepa-framework

Using conda

conda install -c conda-forge jepa-framework

Verification

Verify your installation by running:

python -c "import jepa; print('JEPA installed successfully!')"

Or run the test suite:

python -m pytest tests/

GPU Setup

CUDA Installation

1. Install CUDA Toolkit from NVIDIA website

2. Install cuDNN compatible with your CUDA version

3. Install PyTorch with CUDA support:

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

Verify GPU availability:

python -c "import torch; print(f'CUDA available: {torch.cuda.is_available()}')"

Optional Components

Weights & Biases

1. Create an account at https://wandb.ai

2. Install wandb:

pip install wandb

3. Login:

wandb login

TensorBoard

Install TensorBoard for visualization:

pip install tensorboard

Triton (Advanced)

For kernel optimization:

pip install triton

Development Setup

For contributors and advanced users:

1. Install development dependencies:

pip install -r requirements-dev.txt

2. Install pre-commit hooks:

pre-commit install

3. Run tests:

python -m pytest tests/ -v

Troubleshooting

Common Issues

ImportError: No module named ‘torch’ Install PyTorch: pip install torch

CUDA out of memory Reduce batch size in your configuration

Permission denied errors Use --user flag: pip install --user package_name

Package conflicts Create a fresh virtual environment

Getting Help

• Check the FAQ

• Search GitHub Issues

• Join our Discord

Next Steps

After installation, proceed to the Quick Start Guide to begin using JEPA.