3rd party extensions and code

This page lists various projects that make use of Lasagne or implement additional functionality to be used with the library, such as Layer subclasses and helper functions.

• nolearn: scikit-learn compatible wrappers for neural net libraries, including Lasagne, and other utilities
• craffel's fork: Colin Raffel's fork of Lasagne contains implementations of recurrent and LSTM layers. Merged into the main library in June 2015.
• kaggle-ndsb: winning solution for the National Data Science Bowl on Kaggle (plankton classification)
• lasagne-draw: an implementation of Gregor et al.'s DRAW
• draw_net.py: functions to draw diagrams of a Lasagne network using pydot/graphviz.
• deep_q_rl: an implementation of DeepMind's atari reinforcement learning "deep Q network"
• transformer_network: Spatial Transformer Network implementation in the form of a TransformerLayer class. Merged into the main library in August 2015.
• toppings: A wrapper for automate training of Lasagne networks. Provides additionally a Caffe like interface trough json specification
• Parmesan Parmesan is a library adding variational and semi-supervised neural network models to Lasagne.
• BinaryConnect: training Deep Neural Networks with binary weights during propagations
• TensorNet: implementation of Novikov et al.'s "Tensorizing Neural Networks" by the first author
• NTM-Lasagne: an implementation of Neural Turing Machines, along with reproductions of experiments from the paper.
• neural-doodle: implementation of semantic style transfer
• AgentNet: a library for Deep Reinforcement Learning and custom recurrences