Transformer-based Conditional Variational Autoencoder for Controllable Story Generation
This work addresses the problem of enhancing controllability in story generation for researchers and developers working with large language models, particularly in open-domain long text generation.
This paper explores the use of large-scale latent variable models (LVMs) for neural story generation, focusing on both generation effectiveness and controllability. By integrating latent representation vectors with a Transformer-based pre-trained architecture (GPT2) to create a conditional variational autoencoder (CVAE), the authors demonstrate state-of-the-art conditional generation ability and excellent controllability.
We investigate large-scale latent variable models (LVMs) for neural story generation -- an under-explored application for open-domain long text -- with objectives in two threads: generation effectiveness and controllability. LVMs, especially the variational autoencoder (VAE), have achieved both effective and controllable generation through exploiting flexible distributional latent representations. Recently, Transformers and its variants have achieved remarkable effectiveness without explicit latent representation learning, thus lack satisfying controllability in generation. In this paper, we advocate to revive latent variable modeling, essentially the power of representation learning, in the era of Transformers to enhance controllability without hurting state-of-the-art generation effectiveness. Specifically, we integrate latent representation vectors with a Transformer-based pre-trained architecture to build conditional variational autoencoder (CVAE). Model components such as encoder, decoder and the variational posterior are all built on top of pre-trained language models -- GPT2 specifically in this paper. Experiments demonstrate state-of-the-art conditional generation ability of our model, as well as its excellent representation learning capability and controllability.