The rise of advanced language models has led to widespread adoption of prompt engineering as a technique for enhancing and tailoring these systems. Notably, specialized prompts like Chain-of-Thought have uncovered latent reasoning abilities within these models that were previously unrecognized. Yet, the process of identifying potent prompts has been slow, fueling a need for universal prompt optimization strategies. Regrettably, current prompt learning approaches rarely meet all the criteria for being genuinely “universal”, i.e. automated, discrete, black-box, gradient-free, and interpretable simultaneously. In this thesis, we introduce a novel combination of metaheuristics and black-box prompt learning, bringing forth a series of new prompt-tuning algorithms. Within our paradigm, we dev...[ Read more ]

The rise of advanced language models has led to widespread adoption of prompt engineering as a technique for enhancing and tailoring these systems. Notably, specialized prompts like Chain-of-Thought have uncovered latent reasoning abilities within these models that were previously unrecognized. Yet, the process of identifying potent prompts has been slow, fueling a need for universal prompt optimization strategies. Regrettably, current prompt learning approaches rarely meet all the criteria for being genuinely “universal”, i.e. automated, discrete, black-box, gradient-free, and interpretable simultaneously. In this thesis, we introduce a novel combination of metaheuristics and black-box prompt learning, bringing forth a series of new prompt-tuning algorithms. Within our paradigm, we develop and examine six typical methods: hill climbing, simulated annealing, genetic algorithms with/without crossover, tabu search, and harmony search, demonstrating their effectiveness under white-box and black-box prompt learning settings. In particular, we show that these methods can be employed to discover more human-understandable prompts that were previously unknown in both reasoning and image generation tasks, opening the door to a cornucopia of possibilities in prompt optimization. This enables the optimization of black-box models to be more efficient, allowing diverse task-tailored models to be developed with minimum computational resources.