paper-with-me

Papers

Calibration of Pre-trained Transformers

2020-03-17 · EMNLP 2020 11 · Shrey Desai, Greg Durrett

Pre-trained Transformers are now ubiquitous in natural language processing, but despite their high end-task performance, little is known empirically about whether they are calibrated. Specifically, do these models' posterior probabilities provide an accurate empirical measure of how likely the model is to be correct on a given example? We focus on BERT and RoBERTa in this work, and analyze their calibration across three tasks: natural language inference, paraphrase detection, and commonsense reasoning. For each task, we consider in-domain as well as challenging out-of-domain settings, where models face more examples they should be uncertain about. We show that: (1) when used out-of-the-box, pre-trained models are calibrated in-domain, and compared to baselines, their calibration error out-of-domain can be as much as 3.5x lower; (2) temperature scaling is effective at further reducing calibration error in-domain, and using label smoothing to deliberately increase empirical uncertainty helps calibrate posteriors out-of-domain.

📄 PDF Abstract BibTeX arXiv:2003.07892

Code (1)

shreydesai/calibration 공식 구현 pytorch

Tasks

Natural Language Inference

Methods 이 논문이 사용한 방법론

Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Residual Connection 설명 없음
Attention Dropout Attention Dropout is a type of dropout used in attention-based architectures, where elements are randomly dropped out of the…
Linear Warmup With Linear Decay Linear Warmup With Linear Decay is a learning rate schedule in which we increase the learning rate linearly for $n$ updates and then linearly decay afterwards.
Weight Decay 설명 없음
RoBERTa 설명 없음
Refunds@Expedia|||How do I get a full refund from Expedia? “How do I get a full refund from Expedia? How do I get a full refund from Expedia? – Call ☎️ +1-(888) 829 (0881) or +1-805-330-4056 or +1-805-330-4056 for Quick Help &…
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…

Similar Papers 제목 키워드 기반

Diffusion-Inspired Reconfiguration of Transformers for Uncertainty Calibration

2026-02-09 · Manh Cuong Dao, Quang Hung Pham, Phi Le Nguyen, Thao Nguyen Truong 외 arxiv

Uncertainty calibration in pre-trained transformers is critical for their reliable deployment in risk-sensitive applications. Yet, most existing pre-trained transformers do not have a principled mechanism for uncertainty…

Balancing Accuracy, Calibration, and Efficiency in Active Learning with Vision Transformers Under Label Noise

2025-05-07 · Moseli Mots'oehli, Hope Mogale, Kyungim Baek

Fine-tuning pre-trained convolutional neural networks on ImageNet for downstream tasks is well-established. Still, the impact of model size on the performance of vision transformers in similar scenarios, particularly und…

Active Learning

Bag of Tricks for In-Distribution Calibration of Pretrained Transformers

2023-02-13 · Jaeyoung Kim, Dongbin Na, Sungchul Choi, Sungbin Lim

While pre-trained language models (PLMs) have become a de-facto standard promoting the accuracy of text classification tasks, recent studies find that PLMs often predict over-confidently. Although various calibration met…

ClassificationData AugmentationEnsemble Learningtext-classification+1

BiPC: Bidirectional Probability Calibration for Unsupervised Domain Adaption

2024-09-29 · Wenlve Zhou, Zhiheng Zhou, Junyuan Shang, Chang Niu 외

Unsupervised Domain Adaptation (UDA) leverages a labeled source domain to solve tasks in an unlabeled target domain. While Transformer-based methods have shown promise in UDA, their application is limited to plain Transf…

Domain AdaptationUnsupervised Domain Adaptation

Calibration Attention: Learning Reliability-Aware Representations for Vision Transformers

2025-08-12 · Wenhao Liang, Wei Emma Zhang, Lin Yue, Miao Xu 외 arxiv

Most calibration methods operate at the logit level, implicitly assuming that miscalibration can be corrected without changing the underlying representation. We challenge this assumption and propose \textbf{Calibration A…