Calibrated Language Model Fine-Tuning for In- and Out-of-Distribution Data
This addresses calibration issues for users of fine-tuned language models in text classification, offering incremental improvements over prior methods.
The paper tackles the problem of miscalibration in fine-tuned pre-trained language models for in-distribution and out-of-distribution data by proposing a regularized fine-tuning method with on-manifold and off-manifold regularization, resulting in improved performance over existing calibration methods on six datasets in terms of expectation calibration error, misclassification detection, and OOD detection.
Fine-tuned pre-trained language models can suffer from severe miscalibration for both in-distribution and out-of-distribution (OOD) data due to over-parameterization. To mitigate this issue, we propose a regularized fine-tuning method. Our method introduces two types of regularization for better calibration: (1) On-manifold regularization, which generates pseudo on-manifold samples through interpolation within the data manifold. Augmented training with these pseudo samples imposes a smoothness regularization to improve in-distribution calibration. (2) Off-manifold regularization, which encourages the model to output uniform distributions for pseudo off-manifold samples to address the over-confidence issue for OOD data. Our experiments demonstrate that the proposed method outperforms existing calibration methods for text classification in terms of expectation calibration error, misclassification detection, and OOD detection on six datasets. Our code can be found at https://github.com/Lingkai-Kong/Calibrated-BERT-Fine-Tuning.