Alchemy is a utility and contrib extension to the base TensorFlow library. It includes notebooks and tests for recent research papers and my own experiments.

Getting started

Installing Alchemy requires pip. After that, you’re set.

git clone https://github.com/wenkesj/alchemy
(cd alchemy; pip install -e .)

Check out these notebooks that include detailed examples and implementations of recent research and experiments. All notebooks were created with Google Colab and can be opened and edited freely!:

Contents:

• Data
  • Submodules
  • alchemy.contrib.data.art module
  • Module contents
• Abstract Layers
  • Submodules
  • alchemy.contrib.layers.autoencoder module
  • alchemy.contrib.layers.lateral_impl module
  • alchemy.contrib.layers.wta_impl module
  • Module contents
• Reinforcement Learning
  • Subpackages
  • Submodules
  • alchemy.contrib.rl.core_ops module
  • alchemy.contrib.rl.dataset module
  • alchemy.contrib.rl.experience module
  • alchemy.contrib.rl.serialize module
  • alchemy.contrib.rl.streams module
  • Module contents
• Recurrent Neural Networks
  • Submodules
  • alchemy.contrib.rnn.fast_weights_impl module
  • alchemy.contrib.rnn.stacked_rnn_impl module
  • Module contents
• Training
  • Submodules
  • alchemy.contrib.train.bbb module
  • alchemy.contrib.train.sparse module
  • Module contents
• Multiprocessing
  • Subpackages
  • Submodules
  • alchemy.multiprocessing.pool module
  • Module contents
• Utility
  • Submodules
  • alchemy.utils.array_utils module
  • alchemy.utils.assert_utils module
  • alchemy.utils.data_utils module
  • alchemy.utils.distribution_utils module
  • alchemy.utils.sequence_utils module
  • alchemy.utils.shortcuts module
  • alchemy.utils.type_utils module
  • Module contents