I am currently starting a research project whereby I am trying to convert text of one form into another. i.e. If I were to write a seed sentance of the form "Scientists have finally achieved the ability to induce dreams of electric sheep in the minds of anaesthetized robots" I would like GPT-2 to convert this into "Robots have finally had dreams of electric sheep whilst being anaesthetized by scientists." or some coherent permutation of the underlying structure whereby the main logic of the text is conveyed albeit roughly.
The current open source implementation of GPT-2 seeks to predict the next word, i.e. the seed text is given "Scientist have finally" and the generated text would be " started being paid enough!"
My first presumption was to use some form of GAN, however it became quickly evident that:
Recent work has shown that when both quality and diversity is considered, GAN-generated text is substantially worse than language model generations (Caccia et al., 2018; Tevet et al., 2018; Semeniuta et al., 2018).
How could I most effectively achieve this? Thanks.