(This repository is still under construction. Please let us know if you find any issues :-) )
The framework contains several utility classes to help you to easily build new Conversational Intelligence architectures. The idea is to give you a head start in your NLP/Deep Learning project, removing from you the burden of thinking about how to get the data, or how to develop a simple Deep Learning model. You can start from the examples here and build your own model based on them.
This repository was constructed with a set of NLP/Deep Learning tasks in mind. The datasets and models that we chose to initially support reflect this set of tasks. We intend to keep expanding it to more and more datasets, tasks and models.
How to use this Repository?
This code expects Python 3.4+ and portaudio (as dependency for PyAudio) installed. We recommend you to use a virtual python environment. You can create a new virtual environment with:
# Change directory to the place where you'd like to create a new virtual # environment cd /path/to/where/you/want/to/put/your/environment # Install virtualenv pip install virtualenv # Create the virtual environment with Python 3 virtualenv -p 3 name_of_your_enviroment ## (sometimes one have to provide the full path to python, e.g. on mac: virtualenv -p /usr/local/bin/python3 ) # Activate the virtual environmnent source name_of_your_enviroment/bin/activate
This will create a new folder with the name
the environment set, you will now need to install some dependencies in the
repository. For convenience,
we provide a
requirements.txt file that you can use directly. Just run:
# Clone this repository: git clone https://github.com/mindgarage/Ovation.git # Enter the new folder cd Ovation # Install all the requirements pip install -r requirements.txt sh setup_packages.sh # Tell python where to find the modules export PYTHONPATH=$PYTHONPATH:$PWD
Now you are all set to use the code!
What you will find here?
Below is a small summary of what you can find in this repository. This repository was created with the following pipeline in mind:
- datasets: To build any Deep Learning model, you need data. Datasets
that can be found in the internet come in any format, and it may take
hours for one to reorganize them into the format that is convenient
for him. In the
datasetsfolder, you will find a set of utility classes that simply load the data for you and allow it to be accessed in several ways that we deemed useful for performing the Deep Learning tasks we had in mind.
- models: Now that we have access to the data, we need to write models
that receive the data (in a suitable format) and output some result.
modelsfolder you will find some example model classes that perform tasks such as Named Entity Recognition, Sentiment Classification and Intent Classification.
- templates: Examples of how to use the classes are written in the
Additionally, the following folders have some other useful code:
- tools: Some standalone scripts.
- utils: Utility functions used by the rest of the code.
- tests: Some code used for testing the functionalities above