Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs
This addresses uncertainty estimation for generative LLMs, which is crucial for evaluating reliability, but it is incremental as it focuses on improving scoring functions rather than introducing a new paradigm.
The paper tackles the problem of uncertainty estimation in generative large language models by proposing a trainable scoring function called LARS, which improves reliability and calibration, achieving up to 16% AUROC score gains over existing methods.
Uncertainty estimation (UE) of generative large language models (LLMs) is crucial for evaluating the reliability of generated sequences. A significant subset of UE methods utilize token probabilities to assess uncertainty, aggregating multiple token probabilities into a single UE score using a scoring function. Existing scoring functions for probability-based UE, such as length-normalized scoring and semantic contribution-based weighting, are designed to solve certain aspects of the problem but exhibit limitations, including the inability to handle biased probabilities and complex semantic dependencies between tokens. To address these issues, in this work, we propose Learnable Response Scoring (LARS) function, a novel scoring function that leverages supervised data to capture complex dependencies between tokens and probabilities, thereby producing more reliable and calibrated response scores in computing the uncertainty of LLM generations. Our comprehensive experiments across question-answering and arithmetical reasoning tasks with various datasets demonstrate that LARS significantly outperforms existing scoring functions, achieving improvements of up to 16\% AUROC score.