Multi-hop reasoning, which requires multi-step reasoning based on the supporting documents within a given context, remains challenging for large language models (LLMs). LLMs often struggle to filter out irrelevant documents within the context, and their performance is sensitive to the position of supporting documents within that context. In this paper, we identify an additional challenge: LLMs' performance is also sensitive to the order in which the supporting documents are presented. We refer to this as the misordered context problem. To address this issue, we propose a simple yet effective method called context repetition (CoRe), which involves prompting the model by repeatedly presenting the context to ensure the supporting documents are presented in the optimal order for the model. Using CoRe, we improve the F1 score by up to 30%p on multi-hop QA tasks and increase accuracy by up to 70%p on a synthetic task. Additionally, CoRe helps mitigate the well-known "lost-in-the-middle" problem in LLMs and can be effectively combined with retrieval-based approaches utilizing Chain-of-Thought (CoT) reasoning.

The primary challenge of TTA is limited access to the entire test dataset during online updates, causing error accumulation. To mitigate it, TTA methods have utilized the model output's entropy as a confidence metric that aims to determine which samples have a lower likelihood of causing error. Through experimental studies, however, we observed the unreliability of entropy as a confidence metric for TTA under biased scenarios and theoretically revealed that it stems from the neglect of the influence of latent disentangled factors of data on predictions. Building upon these findings, we introduce a novel TTA method named Destroy Your Object (DeYO), which leverages a newly proposed confidence metric named Pseudo-Label Probability Difference (PLPD). PLPD quantifies the influence of the shape of an object on prediction by measuring the difference between predictions before and after applying an object-destructive transformation. DeYO consists of sample selection and sample weighting, which employ entropy and PLPD concurrently. For robust adaptation, DeYO prioritizes samples that dominantly incorporate shape information when making predictions. Our extensive experiments demonstrate the consistent superiority of DeYO over baseline methods across various scenarios, including biased and wild. Although deep neural networks (DNNs) demonstrate powerful performance across various domains, they lack robustness against distribution shifts under conventional training (He et al., 2016; Pan & Yang, 2009). Therefore, research areas such as domain generalization (Blanchard et al., 2011; Gulrajani & Lopez-Paz, 2021), which involves training models to be robust against arbitrary distribution shifts, and unsupervised domain adaptation (UDA) (Ganin & Lempitsky, 2015; Park et al., 2020), which seeks domain-invariant information for label-absent target domains, have been extensively investigated in the existing literature. Test-time adaptation (TTA) (Wang et al., 2021a) has also gained significant attention as a means to address distribution shifts occurring during test time. TTA leverages each data point once for adaptation immediately after inference. Its minimal overhead compared to existing areas makes it particularly suitable for real-world applications (Azimi et al., 2022). Because UDA assumes access to the entire test samples before adaptation, it utilizes its information on a task by analyzing the distribution of the entire test set (Kang et al., 2019). It leads to inaccurate predictions, and incorporating them into model updates results in error accumulation within the model (Arazo et al., 2020).

TTA guidance from entropy minimization focuses on adapting a model during the inference phase, using loss from misclassified pseudo label. We developed only the test data that is streamed online, without access to a gradient alignment loss to precisely manage the training data or test labels. Common strategies employed in adaptation process, ensuring that changes made for TTA include objectives like entropy minimization [Wang et al., some data don't negatively impact the model's performance 2021] or cross-entropy with pseudo-labels [Goyal et al., 2022], on other data. We introduce a prototype designed to guide the model's self-supervision. However, feature of a class as a proxy measure of the negative these methods are susceptible to confirmation bias [Arazo et impact. To make GAP regularizer feasible under al., 2020], where data with noisy predictions can lead the the TTA constraints, where model can only access model to continually learn in the wrong direction.

The disparity in accuracy between classes in standard training is amplified during adversarial training, a phenomenon termed the robust fairness problem. Existing methodologies aimed to enhance robust fairness by sacrificing the model's performance on easier classes in order to improve its performance on harder ones. However, we observe that under adversarial attacks, the majority of the model's predictions for samples from the worst class are biased towards classes similar to the worst class, rather than towards the easy classes. Through theoretical and empirical analysis, we demonstrate that robust fairness deteriorates as the distance between classes decreases. Motivated by these insights, we introduce the Distance-Aware Fair Adversarial training (DAFA) methodology, which addresses robust fairness by taking into account the similarities between classes. Specifically, our method assigns distinct loss weights and adversarial margins to each class and adjusts them to encourage a trade-off in robustness among similar classes. Experimental results across various datasets demonstrate that our method not only maintains average robust accuracy but also significantly improves the worst robust accuracy, indicating a marked improvement in robust fairness compared to existing methods. Recent studies have revealed the issue of accuracy imbalance among classes (He & Garcia, 2009). This imbalance becomes even more pronounced during adversarial training, which utilizes adversarial examples (Szegedy et al., 2013) to enhance the robustness of the model (Madry et al., 2017). This phenomenon is commonly referred to as "robust fairness problem" (Xu et al., 2021). Existing research has introduced methods inspired by long-tailed (LT) classification studies (He & Garcia, 2009; Zhang et al., 2023) to mitigate the challenge of achieving robust fairness. LT classification tasks tackle the problem of accuracy imbalance among classes, stemming from classifier bias toward classes with a substantial number of samples (head classes) within the LT dataset. The methods proposed for LT classification mainly apply opposing strategies to head classes and tail classes-those classes within LT datasets that have a limited number of samples. For instance, methods proposed by Cao et al. (2019); Khan et al. (2019); Menon et al. (2020) deliberately reduce the model output for head classes while augmenting the output for tail classes by adding constants. These approaches typically lead to improved accuracy for tail classes at the expense of reduced accuracy for head classes. Benz et al. (2021) noted similarities between the fairness issue in LT classification and that in adversarial training. They corresponded the head and tail classes in LT classification with the easy and hard classes in adversarial training, respectively.

Successful detection of Out-of-Distribution (OoD) data is becoming increasingly important to ensure safe deployment of neural networks. One of the main challenges in OoD detection is that neural networks output overconfident predictions on OoD data, make it difficult to determine OoD-ness of data solely based on their predictions. Outlier exposure addresses this issue by introducing an additional loss that encourages low-confidence predictions on OoD data during training. While outlier exposure has shown promising potential in improving OoD detection performance, all previous studies on outlier exposure have been limited to utilizing visual outliers. Drawing inspiration from the recent advancements in vision-language pre-training, this paper venture out to the uncharted territory of textual outlier exposure. First, we uncover the benefits of using textual outliers by replacing real or virtual outliers in the image-domain with textual equivalents. Then, we propose various ways of generating preferable textual outliers. Our extensive experiments demonstrate that generated textual outliers achieve competitive performance on large-scale OoD and hard OoD benchmarks. Furthermore, we conduct empirical analyses of textual outliers to provide primary criteria for designing advantageous textual outliers: near-distribution, descriptiveness, and inclusion of visual semantics.