Calibrated Large Language Models for Binary Question Answering
This work addresses the problem of unreliable uncertainty estimates in LLMs for binary question answering, enhancing interpretability and trustworthiness, though it is incremental as it builds on existing calibration techniques.
The paper tackles the challenge of quantifying uncertainty in large language models (LLMs) for binary text classification by proposing a novel calibration approach using the inductive Venn-Abers predictor (IVAP). The result shows that IVAP consistently outperforms temperature scaling on the BoolQ dataset with Llama 2, achieving well-calibrated probabilities while maintaining high predictive quality.
Quantifying the uncertainty of predictions made by large language models (LLMs) in binary text classification tasks remains a challenge. Calibration, in the context of LLMs, refers to the alignment between the model's predicted probabilities and the actual correctness of its predictions. A well-calibrated model should produce probabilities that accurately reflect the likelihood of its predictions being correct. We propose a novel approach that utilizes the inductive Venn--Abers predictor (IVAP) to calibrate the probabilities associated with the output tokens corresponding to the binary labels. Our experiments on the BoolQ dataset using the Llama 2 model demonstrate that IVAP consistently outperforms the commonly used temperature scaling method for various label token choices, achieving well-calibrated probabilities while maintaining high predictive quality. Our findings contribute to the understanding of calibration techniques for LLMs and provide a practical solution for obtaining reliable uncertainty estimates in binary question answering tasks, enhancing the interpretability and trustworthiness of LLM predictions.