QARAC
In February 2023, the hype and controversy surrounding ChatGPT prompted me to write an article about The Future of NLP, in which I argued that future generations of NLP models will have to incorporate knowledge and a concept of logical consistency.
Subsequently, I decided to start a research project to try to put these ideas into practice. The approach I chose is outlined in QARAC: Question Answering, Reasoning and Consistency. I outlined my architecture, base models, and the data sets I intend to use to train them in QARAC: Models and Corpora. After initially developing the models with Tensorflow, I found it necessary to port them to PyTorch, as described in QARAC: Porting to PyTorch.
Development of QARAC is currently paused until I have the resources to train the models. This page will be updated when that happens.