MEGATRON-CNTRL: Controllable Story Generation with External Knowledge Using Large-Scale Language Models
This work addresses the need for controllable story generation for applications like creative writing or content creation, though it is incremental as it builds on existing large language models by adding external knowledge components.
The paper tackles the problem of uncontrollable text generation in large language models by proposing MEGATRON-CNTRL, a framework that incorporates external knowledge to add control, resulting in more fluent, consistent, and coherent stories with less repetition and higher diversity on the ROC story dataset, and scaling the model to 8.3 billion parameters improved consistency from 74.5% to 93.0% and controllability from 77.5% to 91.5%.
Existing pre-trained large language models have shown unparalleled generative capabilities. However, they are not controllable. In this paper, we propose MEGATRON-CNTRL, a novel framework that uses large-scale language models and adds control to text generation by incorporating an external knowledge base. Our framework consists of a keyword predictor, a knowledge retriever, a contextual knowledge ranker, and a conditional text generator. As we do not have access to ground-truth supervision for the knowledge ranker, we make use of weak supervision from sentence embedding. The empirical results show that our model generates more fluent, consistent, and coherent stories with less repetition and higher diversity compared to prior work on the ROC story dataset. We showcase the controllability of our model by replacing the keywords used to generate stories and re-running the generation process. Human evaluation results show that 77.5% of these stories are successfully controlled by the new keywords. Furthermore, by scaling our model from 124 million to 8.3 billion parameters we demonstrate that larger models improve both the quality of generation (from 74.5% to 93.0% for consistency) and controllability (from 77.5% to 91.5%).