dagisky / cbm_newsqa

Conceptual-Base-Model

Deep neural network for high level reasioning

Table of Contents

1. Project Motivation
2. Installation
3. Usage
4. Licensing and Acknowledgements

Project Motivation

The combinatorial model over sets of high-level representation is a crucial part of
our intelligent behavior thus, to achieve a more natural and complex system in NLU, a higher level of representation is mandatory. The conceptual base theory has an initial premise that the bases of any natural language (NL) is conceptual base a representation that is not necessarily fixed or tied to a single or group of words but rather concepts and conceptualization is regarded as the relationship between these representations.

Installation

The following libraries meke the dependencies for the model

• Python3
• PyTorch
• TensorFlow
• tensorboardX
• matplotlib

Usage

The model works for

Example: The default program runs on multiple gpu's The main.py contains list of GPU's to run the program on.

python main.py # Run on multiple GPU's

However if you want to run on CPU use the use_cuda option

python main.py --use_cuda false # Run on CPU

Note: Eventhough the model can run on cpu, it highly recomanded to use Multiple GPU to train the model.

Licensing and Acknowledgements

UESTC