Adapting LLM predictions in in-context learning with data priors

In-Context Learning (ICL) has enabled Large Language Models (LLMs) to excel as generalpurpose models in zero and few-shot task settings. However, since LLMs are often not trained on the downstream tasks, they lack crucial contextual knowledge from the data distributions, which limits their task adaptability. This paper explores using data priors to automatically customize prompts in ICL. We extract these priors in a dataset-agnostic way based on historical information, enabling LLMs to personalize their output towards users or tasks at inference time. We find that they improve LLM’s output by injecting latent datasetspecific information for the task of rating prediction. Throughout a series of experiments, we show replicable results across LLMs and datasets on what information and methods are most effective for adapting ICL outputs with priors. Our findings offer a systematic approach to customizing prompts with additional information in a privacy-friendly manner, requiring only aggregated data that is computationally efficient.