hippynn

Getting Started:

  • What is hippynn?
  • How to install hippynn
  • Research articles using hippynn
  • License

User Guide:

  • hippynn Features
  • hippynn Concepts
  • Databases
  • Units in hippynn
  • Performance Tips
  • Custom Kernels
  • Model and Loss Graphs
  • Library Settings
  • Creating Custom Node Types

Examples:

  • Minimal Workflow
  • Controller
  • Plotting
  • Predictor
  • Ensembling Models
  • Periodic Boundary Conditions
  • Force Training
  • Restarting training
  • ASE Calculators
  • LAMMPS interface
  • Non-Adiabiatic Excited States
  • Weighted/Masked Loss Functions
  • Pytorch Lightning module
  • Hyperparameter optimization with Ax and Ray

API Documentation:

  • Full API Documentation
  • API Summary Pages
hippynn
  • User Guide
  • View page source

User Guide

Here we explain how the library works.

Contents:

  • hippynn Features
    • Modular set of pytorch layers for atomistic operations
    • Graph level API for simple and flexible construction of models from pytorch components.
    • Plot level API for tracking your training.
    • Training & Experiment API
    • Custom Kernels for fast execution
    • Interfaces
  • hippynn Concepts
    • Layers/Networks
    • Nodes
    • Graphs
    • Experiment
  • Databases
    • Database Formats and notes
  • Units in hippynn
  • Performance Tips
    • Devices
    • Custom Kernels
    • Profiling
    • Scaling
  • Model and Loss Graphs
  • Custom Kernels
    • Bottom line up front
    • Brief Description
    • Comparison Table
    • Detailed Explanation
  • Library Settings
  • Creating Custom Node Types
    • The very basics
    • A MultiNode
    • Parent expansion
    • Adding constraints to possible parents

© Copyright 2019, Los Alamos National Laboratory.

Built with Sphinx using a theme provided by Read the Docs.