DisCup: Discriminator Cooperative Unlikelihood Prompt-tuning for Controllable Text Generation
This addresses the challenge of precise and efficient attribute control in text generation for applications like content creation, though it is incremental as it builds on existing prompt-tuning methods.
The paper tackles the problem of attribute-controllable text generation with large language models, where vanilla prompt tuning suffers from poor generalization and limited control over multiple attributes, and proposes DisCup, which uses a discriminator and unlikelihood objective to optimize prompts, achieving state-of-the-art control performance with only around 10 virtual tokens.
Prompt learning with immensely large Casual Language Models (CLMs) has been shown promising for attribute-controllable text generation (CTG). However, vanilla prompt tuning tends to imitate training corpus characteristics beyond the control attributes, resulting in a poor generalization ability. Moreover, it is less able to capture the relationship between different attributes, further limiting the control performance. In this paper, we propose a new CTG approach, namely DisCup, which incorporates the attribute knowledge of discriminator to optimize the control-prompts, steering a frozen CLM to produce attribute-specific texts. Specifically, the frozen CLM model, capable of producing multitudinous texts, is first used to generate the next-token candidates based on the context, so as to ensure the diversity of tokens to be predicted. Then, we leverage an attribute-discriminator to select desired/undesired tokens from those candidates, providing the inter-attribute knowledge. Finally, we bridge the above two traits by an unlikelihood objective for prompt-tuning. Extensive experimental results show that DisCup can achieve a new state-of-the-art control performance while maintaining an efficient and high-quality text generation, only relying on around 10 virtual tokens.