rnn : Recurrent Library for Torch

Abstract

The rnn package provides components for implementing a wide range of Recurrent Neural Networks. It is built withing the framework of the Torch distribution for use with the nn package. The components have evolved from 3 iterations, each adding to the flexibility and capability of the package. All component modules inherit either the AbstractRecurrent or AbstractSequencer classes. Strong unit testing, continued backwards compatibility and access to supporting material are the principles followed during its development. The package is compared against existing implementations of two published papers.

References in corpus (3)

• Improving neural networks by preventing co-adaptation of feature detectors
• Recurrent Neural Network Regularization
• Recurrent Models of Visual Attention

Cited by in corpus (8)

• Fine-grained Analysis of Sentence Embeddings Using Auxiliary Prediction Tasks
• Multimodal Residual Learning for Visual QA
• Combining Fully Convolutional and Recurrent Neural Networks for 3D Biomedical Image Segmentation
• Optimizing Performance of Recurrent Neural Networks on GPUs
• Learning Global Features for Coreference Resolution
• Comparing Neural- and N-Gram-Based Language Models for Word Segmentation
• LSTMVis: A Tool for Visual Analysis of Hidden State Dynamics in Recurrent Neural Networks
• Learning Visual Knowledge Memory Networks for Visual Question Answering