TransPoemer
Transforming regular text to a poem (limerick) via the means of a Transformer model (neural network). The figure below shows the general model overview. The model consists out of two steps:
- Extract the keywords from the text on which the text is based on
- Translate these keywords to a poem
Aside from the task to translate regular text into a poem, this repository investigates if it's possible to train such a network on a laptop, requiring significantly less computational power than related state-of-the-art Transformer models.
Report and Presentation
This project was part of a school project. The report as well as the final presentation can be found in the root
directory; TransPoemer_report.pdf and TransPoemer_presentation.pptx respectively.
Project
All functionality implemented in this project is handled via main.py (root directory).