Paper · arXiv 2406.04370
Large Language Model Confidence Estimation via Black-Box Access
Tejaswini Pedapati, Amit Dhurandhar, Soumya Ghosh, Soham Dan, Prasanna Sattigeri
A framework using engineered features and logistic regression effectively estimates the confidence of large language models with black-box access, outperforming existing methods on multiple datasets.
Abstract
Estimating uncertainty or confidence in the responses of a model can be significant in evaluating trust not only in the responses, but also in the model as a whole. In this paper, we explore the problem of estimating confidence for responses of large language models (LLMs) with simply black-box or query access to them. We propose a simple and extensible framework where, we engineer novel features and train a (interpretable) model (viz. logistic regression) on these features to estimate the confidence. We empirically demonstrate that our simple framework is effective in estimating confidence of flan-ul2, llama-13b and mistral-7b with it consistently outperforming existing black-box confidence estimation approaches on benchmark datasets such as TriviaQA, SQuAD, CoQA and Natural Questions by even over 10% (on AUROC) in some cases. Additionally, our interpretable approach provides insight into features that are predictive of confidence, leading to the interesting and useful discovery that our confidence models built for one LLM generalize zero-shot across others on a given dataset.