dlutils

First steps

  • dlutils
  • Batch provider - for parallel batch data processing
  • Checkpointer - saving/restoring of model/optimizers/schedulers/custom data
  • Run - helper for launching distributed parallel training
  • Download - module for downloading and unpacking files
  • Shuffle functions for ndarrays.
  • Decorator for measuring time
  • Utils for organizing epoch iterations, printing progress, computing average of losses.
  • Measures
  • Caching decorator
  • Async decorator
  • Processing a large tensor by chunking
  • Readers
dlutils
  • Docs »
  • Overview: module code

All modules for which code is available

  • dlutils.async
  • dlutils.batch_provider
  • dlutils.block_process
  • dlutils.cache
  • dlutils.default_cfg
  • dlutils.download
  • dlutils.epoch
  • dlutils.measures
  • dlutils.numpy_dataset
  • dlutils.progress_bar
  • dlutils.pytorch.checkpointer
  • dlutils.pytorch.jacobian
  • dlutils.pytorch.launcher
  • dlutils.random_rotation
  • dlutils.reader
  • dlutils.registry
  • dlutils.save_image
  • dlutils.shuffle
  • dlutils.timer
  • dlutils.tracker

© Copyright 2019-2020, Stanislav Pidhorskyi

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