Ignite Your Networks!

PyTorch-Ignite is a high-level library to help with training and evaluating neural networks in PyTorch flexibly and transparently.

All our documentation moved to pytorch-ignite.ai

Package Reference

• ignite.engine
  • ignite.engine.engine
  • ignite.engine.events
  • ignite.engine.deterministic
  • helper methods to define supervised trainer and evaluator
  • Resuming the training
  • Deterministic training
  • Dataflow synchronization
• ignite.handlers
  • Complete list of generic handlers
  • Loggers
  • Parameter scheduler
  • State Parameter scheduler
  • More on parameter scheduling
• ignite.metrics
  • Attach Engine API
  • Reset, Update, Compute API
  • Metric arithmetics
  • How to create a custom metric
  • Metrics and its usages
  • Metrics and distributed computations
  • Complete list of metrics
  • Helpers for customizing metrics
• ignite.distributed
  • Distributed launcher and auto helpers
  • ignite.distributed.auto
  • ignite.distributed.launcher
  • ignite.distributed.utils
• ignite.exceptions
  • NotComputableError
• ignite.utils
  • apply_to_tensor()
  • apply_to_type()
  • convert_tensor()
  • hash_checkpoint()
  • manual_seed()
  • setup_logger()
  • to_onehot()

Contrib Package Reference

• ignite.contrib.engines
  • Truncated Backpropagation Throught Time
  • Helper methods to setup trainer/evaluator
• ignite.contrib.metrics
  • Contrib module metrics [deprecated]
  • Regression metrics [deprecated]
• ignite.contrib.handlers
  • Parameter scheduler [deprecated]
  • LR finder [deprecated]
  • Time profilers [deprecated]
  • Loggers [deprecated]