Introduction to PyTorch on SeaWulf

PyTorch is an open-source machine learning framework widely used for deep learning research and production. It provides flexibility and speed when building complex neural networks, especially in research areas such as computer vision and natural language processing.

SeaWulf offers various versions of PyTorch, supporting both CPU and GPU configurations to accommodate different computational needs and hardware preferences.

Available PyTorch Versions

SeaWulf provides the following versions of PyTorch, both CPU and GPU optimized:

CPU Versions

• pytorch-cpu/latest (latest CPU version)
• pytorch-latest/1.13.0
• pytorch/1.4.0
• pytorch/0.3.0

GPU Versions

• pytorch-gpu/2.0.1 (latest GPU version)
• pytorch-gpu/latest (latest available GPU version)

PyTorch Lightning Version

• pytorch-lightning/2.1.3 (for easier training and scaling of PyTorch models)

To load a specific version of PyTorch, use the module load command: