Manifestro Docs

Installation

Install DREAM and get ready to use adaptive neural dynamics

Installation

This guide covers installing DREAM (Dynamic Recall and Elastic Adaptive Memory) in your Python environment.

Requirements

Before installing DREAM, ensure you have:

  • Python ≥ 3.9
  • PyTorch ≥ 2.0.0
  • CUDA (optional, for GPU acceleration)

Installation Methods

The easiest way to install DREAM is via PyPI:

pip install dreamnn

From Source

For the latest development version:

git clone https://github.com/karl4th/dream-nn.git
cd dream-nn
pip install -e .

Development Installation

If you want to contribute or access development tools:

pip install -e ".[dev]"

This includes:

  • Testing frameworks
  • Code quality tools
  • Development dependencies

Verify Installation

After installation, verify that DREAM is working correctly:

import torch
from dream import DREAM, DREAMConfig

print(f"PyTorch version: {torch.__version__}")
print(f"DREAM available: {DREAM is not None}")

# Quick test
model = DREAM(input_dim=10, hidden_dim=32)
x = torch.randn(2, 5, 10)
output, state = model(x)
print(f"Test passed! Output shape: {output.shape}")

Expected output:

PyTorch version: 2.x.x
DREAM available: True
Test passed! Output shape: torch.Size([2, 5, 32])

Platform-Specific Instructions

Linux

pip install dreamnn

For GPU support, install PyTorch with CUDA:

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

macOS

pip install dreamnn

For Apple Silicon (M1/M2) with MPS support:

pip install torch torchvision torchaudio
pip install dreamnn

Windows

pip install dreamnn

For GPU support on Windows:

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

Troubleshooting

Import Error: No module named 'dream'

Ensure you're in the correct Python environment:

python -m pip install dreamnn

Or check your Python path:

python -c "import sys; print(sys.executable)"

CUDA Not Available

If you want GPU acceleration but CUDA isn't available:

  1. Check NVIDIA driver version: nvidia-smi
  2. Install matching CUDA version
  3. Reinstall PyTorch with correct CUDA version

Version Conflicts

If you encounter version conflicts:

pip install --upgrade pip
pip install dreamnn --force-reinstall

Next Steps

Now that DREAM is installed, check out the Quick Start guide to learn the basics!

On this page