NeMo: Neural Modules Toolkit

Neural Modules (NeMo): a framework-agnostic toolkit for building AI applications powered by Neural Modules

VIDEO

A Short VIDEO walk-through about using NEMO to experiment with ASR systems.

Core Concepts and Features

• NeuralModule class - represents and implements a neural module.
• NmTensor - represents activation flow between neural modules' ports.
• NeuralType - represents types of modules' ports and NmTensors.
• NeuralFactory - to create neural modules and manage training.
• Lazy execution - when describing activation flow between neural modules, nothing happens until an "action" (such as optimizer.optimize(...) is called.
• Collections - NEMO comes with collections - related group of modules such as nemo_asr (for Speech Recognition) and nemo_nlp for NLP

Documentation

Please refer to the HTML documentation in the docs folder

Requirements

1. Python 3.6 or 3.7
2. Pytorch >=1.0 with GPU support
3. NVIDIA APEX: https://github.com/NVIDIA/apex
4. (for nemo_asr do: apt-get install libsndfile1)

Unittests

Run this:

./reinstall.sh
python -m unittest tests/*.py

Getting started

1. Clone the repository and run unittests
2. Go to nemo folder and do: python setup.py install

3. Install collections:

   1. ASR collection from collections/nemo_asr do: python setup.py install
   2. NLP collection coming soon ...

4. For development do: python setup.py develop instead of python setup.py install in Step (3) above
5. Go to examples/start_here to get started with few simple examples

6. To get started with speech recognition:

   1. head to the ASR tutorial in the documentation
   2. head to examples/asr/ASR_made_simple.ipynb