Quickstart

Requirements

Operating System

isciml is designed to work on Unix-based systems. We recommend using:

• Ubuntu 20.04 LTS or newer

• CentOS 8 or newer

• macOS 10.15 (Catalina) or newer

Windows users should consider using Windows Subsystem for Linux (WSL 2) with Ubuntu.

Software Requirements

• Docker 20.10 or newer

• Singularity 3.5 or newer

• Python 3.8 or newer (for any local processing or scripting)

• CUDA-capable GPU for optimal performance (CUDA 11.0 or newer)

Hardware Recommendations

• CPU: 4+ cores

• RAM: 16GB+

• Storage: 50GB+ free space

• GPU: NVIDIA GPU with at least 8GB VRAM (e.g., NVIDIA RTX 2070 or better)

Installation

Install Docker and Singularity

Before installing isciml, you need to have Docker and Singularity installed on your system. Please follow the official installation guides for your operating system:

• Docker: https://docs.docker.com/get-docker/

• Singularity: https://sylabs.io/guides/3.0/user-guide/installation.html

Note: Singularity doesn't have native support for macOS. For macOS users, we recommend using a virtual machine with Linux or Docker Desktop with a Linux container.

Install isciml

Once Docker and Singularity are installed, you can proceed with installing isciml:

1. Pull the latest isciml container image:

singularity pull --name isciml.sif docker://s2labs/isciml:latest

This command will create a Singularity Image File (SIF) named isciml.sif in your current directory.

1. Verify the installation by running:

singularity run isciml.sif isciml --help

This should display the help information for isciml.

First Steps

After installation, you can start using isciml. Here's a basic workflow:

1. Generate models:

singularity exec isciml.sif isciml generate-models [OPTIONS]

1. Generate data:

singularity exec isciml.sif isciml generate [OPTIONS]

1. Train a model:

singularity exec isciml.sif isciml train [OPTIONS]

1. Run inference:

singularity exec isciml.sif isciml inference [OPTIONS]

Replace [OPTIONS] with the appropriate options for your use case. Refer to the documentation for each subcommand for detailed information on available options.

Tips

• Regularly check for updates to the isciml container and update your local image:

  singularity pull --name isciml.sif docker://s2labs/isciml:latest

• When working with large datasets or complex models, consider using a machine with more RAM and a powerful GPU.

• For production environments, consider setting up a dedicated server or using cloud computing resources with GPU support.

Troubleshooting

If you encounter any issues:

1. Ensure your Docker and Singularity installations are up to date.

2. Check that your system meets the minimum requirements.

3. Verify that you're using the correct syntax for Singularity commands.

4. Consult the isciml documentation or reach out to the support channels provided by the isciml developers.

For more detailed information on using isciml, refer to the full documentation for each subcommand.