To all Python installations in the module system the following packages have been added:
- TensorFlow, the arguably most used machine learning framework (https://www.tensorflow.org/)
- PyTorch, another ML framework with focus on Deep Learning (https://pytorch.org/)
- Keras, a Deep Learning library with interface to TensorFlow (https://keras.io/)
- h5py, a Python interface for the HDF5 data format (http://www.h5py.org/)
- mpi4py, a MPI binding for Python, to allow computing with multiple nodes (https://pypi.org/project/mpi4py/)
Additionally, Lichtenberg cluster now also supports Singularity containers (https://singularity.lbl.gov/). This facilitates external development of programs and workflows (for example on one's own notebook), and then executing it on the Lichtenberg cluster, without having to look after dependencies or differences in development environments.
For more information and consultation regarding the usage of the modules, please contact the Lichtenberg team. (email@example.com-…).