Vision Transformers (ViTs) have shown success across a variety of tasks due to their ability to capture global image representations. Recent studies have identified the existence of high-norm tokens in ViTs, which can interfere with unsupervised object discovery. To address this, the use of "registers" which are additional tokens that isolate high norm patch tokens while capturing global image-level information has been proposed. While registers have been studied extensively for object discovery, their generalization properties particularly in out-of-distribution (OOD) scenarios, remains underexplored. In this paper, we examine the utility of register token embeddings in providing additional features for improving generalization and anomaly rejection. To that end, we propose a simple method that combines the special CLS token embedding commonly employed in ViTs with the average-pooled register embeddings to create feature representations which are subsequently used for training a downstream classifier. We find that this enhances OOD generalization and anomaly rejection, while maintaining in-distribution (ID) performance. Extensive experiments across multiple ViT backbones trained with and without registers reveal consistent improvements of 2-4% in top-1 OOD accuracy and a 2-3% reduction in false positive rates for anomaly detection. Importantly, these gains are achieved without additional computational overhead.

Anchoring is a recent, architecture-agnostic principle for training deep neural networks that has been shown to significantly improve uncertainty estimation, calibration, and extrapolation capabilities. In this paper, we systematically explore anchoring as a general protocol for training vision models, providing fundamental insights into its training and inference processes and their implications for generalization and safety. Despite its promise, we identify a critical problem in anchored training that can lead to an increased risk of learning undesirable shortcuts, thereby limiting its generalization capabilities. To address this, we introduce a new anchored training protocol that employs a simple regularizer to mitigate this issue and significantly enhances generalization. We empirically evaluate our proposed approach across datasets and architectures of varying scales and complexities, demonstrating substantial performance gains in generalization and safety metrics compared to the standard training protocol. The open-source code is available at https://software.llnl.gov/anchoring.

We address the problem of predicting the generalization gap of deep neural networks under large, natural, and synthetic distribution shifts between source and target domains. This is crucial in understanding how models behave in uncontrollable ‘in-the-wild’ scenarios, but existing techniques fail when target domain becomes very different from the source. Accurately capturing the relationship and distance between the source and target domains is critical for a reliable post-hoc estimation of generalization. In this paper, we propose a novel strategy for directly predicting accuracy on unseen target data with the help of anchoring and pre-text encoding in predictive models. Anchoring has been shown previously to perform effectively in characterizing domain shifts, which we exploit for predicting the generalization gap. Our experiments on the PACS dataset along with synthetic ablations indicate that our approach produces well calibrated accuracy estimates outperforming existing baselines.