Few-Shot NLG with Pre-Trained Language Model
This addresses the difficulty of adopting NLG for real-world applications with limited data, though it is incremental as it builds on existing pre-trained language models.
The paper tackles the problem of data-hungry neural natural language generation (NLG) by proposing a few-shot NLG task, achieving an average improvement of over 8.0 BLEU points with only 200 training examples across domains.
Neural-based end-to-end approaches to natural language generation (NLG) from structured data or knowledge are data-hungry, making their adoption for real-world applications difficult with limited data. In this work, we propose the new task of \textit{few-shot natural language generation}. Motivated by how humans tend to summarize tabular data, we propose a simple yet effective approach and show that it not only demonstrates strong performance but also provides good generalization across domains. The design of the model architecture is based on two aspects: content selection from input data and language modeling to compose coherent sentences, which can be acquired from prior knowledge. With just 200 training examples, across multiple domains, we show that our approach achieves very reasonable performances and outperforms the strongest baseline by an average of over 8.0 BLEU points improvement. Our code and data can be found at \url{https://github.com/czyssrs/Few-Shot-NLG}