Controllable Generation from Pre-trained Language Models via Inverse Prompting

By Xu Zou, Da Yin, Qingyang Zhong, Hongxia Yang, Zhilin Yang, Jie Tang. 2021

66 citations recorded by the source corpus.

Large-scale pre-trained language models have demonstrated strong capabilities of generating realistic texts. However, it remains challenging to control the generation results. Previous approaches such as prompting are far from sufficient, and lack of controllability limits the usage of language models. To tackle this challenge, we propose an innovative method, inverse prompting, to better control text generation. The core idea of inverse prompting is to use generated text to inversely predict the prompt during beam search, which enhances the relevance between the prompt and the generated text and thus improves controllability. Empirically, we pre-train a large-scale Chinese language model to perform a systematic study using human evaluation on the tasks of open-domain poem generation and open-domain long-form question answering. Results demonstrate that our proposed method substantially outperforms the baselines and that our generation quality is close to human performance on some of the tasks.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Large Language Models, Prompting & Reasoning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Large Language Models

In Prompting & Reasoning

Citations in the corpus

CogView: Mastering Text-to-Image Generation via Transformers (2021, methodology): “CapLoss ( x, t ) is the cross-entropy loss for the text tokens, and this method can be seen as an adaptation of inverse prompting [56] for text-to-image generation.”
Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learning (2022, methodology): “Prompt tuning has been applied to various applications for NLP including continual learning [59], model robustness [60, 61], summarization [62], machine translation [63], co-training [64], probing language models [65, 65], inverse prompting [66] and transfer learning [67].”
Pre-Trained Models: Past, Present and Future (2021, background): “Methods such as InversePrompting (Zou et al., 2021) which shows supreme ability in controlling theme-related text generation would provide some thoughts.”
RLPrompt: Optimizing Discrete Text Prompts with Reinforcement Learning (2022, background): “Complementary to the works above which focus on finding prompts, Zou et al. (2021) augment the generation decoding objective using the prompt, leading to improved performance in poetry generation and long-form QA.”
Learning to Imagine: Integrating Counterfactual Thinking in Neural Discrete Reasoning (2022, background): “However, such uncontrollable model (Zou et al., 2021) can hardly generate high-quality context for two reasons: 1) the context is more complex than plain text, which can include a table (Figure 1); and 2) NDR requires a precise context with the correct numbers (Figure 1, $132,935 for the ﬁnished…”
A Survey of Controllable Text Generation Using Transformer-based Pre-trained Language Models (2022, methodology): “In order to tackle the problem that the distance between the prompt and the next predicted to-ken correlates negatively with the prompt’s influence power, Zou et al. [172] propose a method called Inverse Prompt.”