Improving Structural Diversity of Blackbox LLMs via Chain-of-Specification Prompting
This addresses the problem of limited user control over diversity dimensions in LLM outputs for domains like poetry and code, offering a novel prompting method for blackbox models.
The paper tackles the challenge of controlling diversity in text generation from large language models by introducing structural diversity, a user-defined metric based on specific features like rhyme in poetry or expression types in code, and proposes chain-of-specification prompting to enhance it, showing significant improvements over baselines in experiments.
The capability to generate diverse text is a key challenge facing large language models (LLMs). Thus far, diversity has been studied via metrics such as $n$-gram diversity or diversity of BERT embeddings. However, for these kinds of diversity, the user has little control over the dimensions along which diversity is considered. For example, in the poetry domain, one might desire diversity in terms of rhyme and meter, whereas in the code domain, one might desire diversity in terms of the kinds of expressions used to solve a problem. We propose a diversity metric called structural diversity, where the user provides a mapping from generated text to features capturing the kinds of diversity that they care about. In addition, we propose a novel strategy called chain-of-specification (CoS) prompting for improving diversity by first having the LLM generate a specification encoding one instance of structural features, and then prompting the LLM to generate text that satisfies these features; notably, our strategy works with blackbox LLMs. In our experiments, we show that for structural diversity in the poetry and code domains, CoS significantly improves diversity compared to several baselines.