PiNN: a Python library for building atomic neural networks

• Introduction
  • Flexibility
  • Scalability
  • Examples
• Datasets
  • Loading data
  • Manipulating datasets
  • Implemented loaders
• Networks
  • Structures of ANN
  • PiNN’s representation of ANNs
  • Implemented networks
• Models
  • Write a model
  • Implemented models
• ASE interface
• Utilities
  • PiNNboard
  • Cloud/CLI trainer
  • Use PiNN with Ray Tune
• Notebooks
  • Quick tour with QM9
  • Optimizing the training
  • Customizing dataset
  • Writing an atomic neural network
  • Learning a LJ potential
  • Visualizing Ray Tune result
• Reference
• Miscellaneous Notes
  • Known issue with ASE barostats