Topical Language Generation using Transformers
This addresses the challenge of generating text with specific topical properties for users needing controlled language generation, representing an incremental improvement over existing methods.
The paper tackles the problem of controlling topic, style, and sentiment in text generation from large language models by proposing a Bayesian approach that combines a pre-trained LM with topic modeling, outperforming state-of-the-art methods in coherency, diversity, and fluency with faster decoding.
Large-scale transformer-based language models (LMs) demonstrate impressive capabilities in open text generation. However, controlling the generated text's properties such as the topic, style, and sentiment is challenging and often requires significant changes to the model architecture or retraining and fine-tuning the model on new supervised data. This paper presents a novel approach for Topical Language Generation (TLG) by combining a pre-trained LM with topic modeling information. We cast the problem using Bayesian probability formulation with topic probabilities as a prior, LM probabilities as the likelihood, and topical language generation probability as the posterior. In learning the model, we derive the topic probability distribution from the user-provided document's natural structure. Furthermore, we extend our model by introducing new parameters and functions to influence the quantity of the topical features presented in the generated text. This feature would allow us to easily control the topical properties of the generated text. Our experimental results demonstrate that our model outperforms the state-of-the-art results on coherency, diversity, and fluency while being faster in decoding.