TY - UNPB
T1 - Do Not Design, Learn
T2 - A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs
AU - Yaldiz, Duygu Nur
AU - Bakman, Yavuz Faruk
AU - Buyukates, Baturalp
AU - Tao, Chenyang
AU - Ramakrishna, Anil
AU - Dimitriadis, Dimitrios
AU - Avestimehr, Salman
PY - 2024/6/17
Y1 - 2024/6/17
N2 - In this work, we introduce the Learnable Response Scoring Function (LARS) for Uncertainty Estimation (UE) in generative Large Language Models (LLMs). Current scoring functions for probability-based UE, such as length-normalized scoring and semantic contribution-based weighting, are designed to solve specific aspects of the problem but exhibit limitations, including the inability to handle biased probabilities and under-performance in low-resource languages like Turkish. To address these issues, we propose LARS, a 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 generations. Our extensive experiments across multiple datasets show that LARS substantially outperforms existing scoring functions considering various probability-based UE methods.
AB - In this work, we introduce the Learnable Response Scoring Function (LARS) for Uncertainty Estimation (UE) in generative Large Language Models (LLMs). Current scoring functions for probability-based UE, such as length-normalized scoring and semantic contribution-based weighting, are designed to solve specific aspects of the problem but exhibit limitations, including the inability to handle biased probabilities and under-performance in low-resource languages like Turkish. To address these issues, we propose LARS, a 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 generations. Our extensive experiments across multiple datasets show that LARS substantially outperforms existing scoring functions considering various probability-based UE methods.
KW - cs.CL
U2 - 10.48550/arXiv.2406.11278
DO - 10.48550/arXiv.2406.11278
M3 - Preprint
SP - 1
EP - 16
BT - Do Not Design, Learn
PB - arXiv
ER -