Skip to content

CryoSiam Installation Guide

This guide provides instructions for installing CryoSiam using either pip or conda, with full support for GPU-enabled PyTorch.

Requirements

  • Python >= 3.8
  • Conda (optional, for environment-based setup)
  • (Optional) NVIDIA GPU with CUDA 11.8+ drivers for GPU acceleration

Option 1: Install via Conda Environment

You can also use Conda to create an isolated environment with all dependencies.

Step 1: Create the Environment

conda env create -f https://github.com/frosinastojanovska/cryosiam/blob/main/environment.yml
conda activate cryosiam

Step 2: Install cryosiam

After activating the environment:

git clone https://github.com/frosinastojanovska/cryosiam.git
cd cryosiam
pip install --no-deps .

Option 2: Install via pip

Step 1: Install GPU-enabled PyTorch

Before installing this package, install the correct PyTorch build with GPU support:

pip install torch==2.1.2 torchvision==0.16.2 --index-url https://download.pytorch.org/whl/cu118

If you don’t need GPU support, you can install the CPU-only version instead:

pip install torch==2.1.2 torchvision==0.16.2 --index-url https://download.pytorch.org/whl/cpu

Step 2: Install the Package and Dependencies

Clone the repository (if applicable) and install:

git clone https://github.com/frosinastojanovska/cryosiam.git
cd cryosiam

# Recommended: use a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install the package and dependencies
pip install -r requirements.txt
pip install --no-deps .

Verify Installation

To verify the CryoSiam installation, run:

cryosiam --version

To check that PyTorch is installed correctly and can detect your GPU:

import torch
print(torch.cuda.is_available())  # Should return True if GPU is available

Troubleshooting

  • CUDA errors: Make sure your system has the correct NVIDIA drivers and CUDA version.
  • Missing packages: Double-check that you followed the correct pip or conda install steps.
  • Conflicts: It's recommended to use a clean virtual environment or conda environment.

Need Help?

If you run into issues, feel free to open an issue on the GitHub repository.

Don't Have Conda Installed?

If you don't have conda installed yet, we recommend using Miniforge (lightweight) or Anaconda (full-featured).

Miniforge is a minimal installer for Conda that supports conda-forge by default.

Linux / macOS:

# Download and install Miniforge
wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh
bash Miniforge3-Linux-x86_64.sh

Windows:

  1. Download the latest installer from:
    https://github.com/conda-forge/miniforge/releases

  2. Run the .exe installer and follow the setup.

Option 2: Install Anaconda (Full Distribution)

Anaconda includes Conda, Python, and hundreds of data science packages.

Download from: https://www.anaconda.com/products/distribution

After installing Miniforge or Anaconda, return to the main Conda installation steps above.