Goal-Directed Story Generation: Augmenting Generative Language Models with Reinforcement Learning

Abstract: The advent of large pre-trained generative LLMs has provided a common framework for AI story generation via sampling the model to create sequences that continue the story. However, sampling alone is insufficient for story generation. In particular, it is hard to direct a LLM to create stories to reach a specific goal event. We present two automated techniques grounded in deep reinforcement learning and reward shaping to control the plot of computer-generated stories. The first utilizes proximal policy optimization to fine-tune an existing transformer-based LLM to generate text continuations but also be goal-seeking. The second extracts a knowledge graph from the unfolding story, which is used by a policy network with graph attention to select a candidate continuation generated by a LLM. We report on automated metrics pertaining to how often stories achieve a given goal event as well as human participant rankings of coherence and overall story quality compared to baselines and ablations.