RETURNN: The RWTH Extensible Training framework for Universal Recurrent Neural Networks

2016/08/02 by Patrick Doetsch, Albert Zeyer, Doetsch, Patrick +9 · 1 voice · 2 citations
Computer Science · #Neural Networks and Applications #Speech Recognition and Synthesis #Topic Modeling #cs.CL #cs.LG #cs.NE

paper · pdf · doi:10.48550/arxiv.1608.00895

arxiv published 2016/08/02 · arxiv updated 2017/01/10

Abstract

In this work we release our extensible and easily configurable neural network training software. It provides a rich set of functional layers with a particular focus on efficient training of recurrent neural network topologies on multiple GPUs. The source of the software package is public and freely available for academic research purposes and can be used as a framework or as a standalone tool which supports a flexible configuration. The software allows to train state-of-the-art deep bidirectional long short-term memory (LSTM) models on both one dimensional data like speech or two dimensional data like handwritten text and was used to develop successful submission systems in several evaluation campaigns.

Cited by

• Analysis of Domain Shift across ASR Architectures via TTS-Enabled Separation of Target Domain and Acoustic Conditions
• A Comparative Analysis on ASR System Combination for Attention, CTC, Factored Hybrid, and Transducer Models

Discussions

• RETURNN: The RWTH Extensible Training Framework for Universal RNN [hn, 3 points, 0 comments]

Related

• Scaling Recurrent Neural Network Language Models
• Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) network
• A Review of Recurrent Neural Networks: LSTM Cells and Network Architectures
• From Nodes to Networks: Evolving Recurrent Neural Networks
• Echo: Compiler-based GPU Memory Footprint Reduction for LSTM RNN Training