发表机构
Scuola Internazionale Superiore di Studi Avanzati (SISSA); Flatiron Institute（国际高等研究学院（SISSA）; 弗拉蒂伦研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对提示词优化任务的不适定性，本文提出BayesPrompt算法，将提示词优化重定义为贝叶斯后验推断以生成可读提示词，在真实数据集上较当前最优方法有显著指标提升。
AI 中文摘要
重构能促使大语言模型（LLM）生成期望答案或行为的提示词是一个开放且重要的研究课题。然而，旨在最小化给定答案困惑度的优化方法，始终会产生所谓的伪提示词——无法理解的词元字符串，缺乏人类可解释性。我们认为这是提示词优化任务不适定性的结果。通过将该任务重新定义为对提示词的贝叶斯后验推断，我们提出一种高效算法，可采样出既高效（困惑度方面）又具有人类可读性的提示词。我们将我们的方法与当前最优替代方案在真实数据集上对比，结果显示在一系列指标上有显著提升。
英文摘要
Reconstructing prompts that can elicit a desired answer or behaviour in an LLM is an open and important research topic. Optimisation methods which aim at minimising the perplexity of a given answer, however, consistently yield so-called pseudoprompts, unintelligible strings of tokens which can lack human interpretability. We argue that this is a consequence of the ill-posedness of the prompt optimisation task. By reframing the task as a Bayesian posterior inference over prompts, we propose an efficient algorithm to sample prompts which are both efficient (in terms of perplexity) and human readable. We compare our approach with state of the art alternatives showing on a real data set a marked improvement over a range of metrics.
