Thank You BART! Rewarding Pre-trained Models Improves Formality Style Transfer

Huiyuan Lai, Antonio Toral, Malvina Nissim. Arxiv 2021 – 15 citations

[Paper]    
GPT Training Techniques Fine-Tuning Reinforcement Learning Model Architecture

Scarcity of parallel data causes formality style transfer models to have scarce success in preserving content. We show that fine-tuning pre-trained language (GPT-2) and sequence-to-sequence (BART) models boosts content preservation, and that this is possible even with limited amounts of parallel data. Augmenting these models with rewards that target style and content – the two core aspects of the task – we achieve a new state-of-the-art.

Similar Work