GTC Silicon Valley-2019 ID:S9838:Neural Monkey: A Natural Language Processing Toolkit
Jind?ich Helcl(Charles University)
We'll describe Neural Monkey, an open-source toolkit for sequence-to-sequence learning, which is designed for students and researchers with an emphasis on modularity and code readability. Our goal is for the toolkit to be a growing collection of deep learning models and techniques. We'll explain how Neural Monkey's simple configuration files make it relatively easy to create and experiment with new model architectures. The toolkit, implemented in Python 3.5 using the TensorFlow library, provides a straightforward way to run computations on GPUs. We'll also discuss how the toolkit's modular design makes it easily extensible, with easy access to the created component from the configuration file.