Ensuring that deep learning models are well-calibrated in terms of their predictive uncertainty is essential in maintaining their trustworthiness and reliability, yet despite increasing advances in foundation model research, the relationship between such large language models (LLMs) and their calibration remains an open area of research. In this work, we look at a critical gap in the calibration of LLMs within multilingual settings, in an attempt to better understand how the data scarcity can potentially lead to different calibration effects and how commonly used techniques can apply in these settings. Our analysis on two multilingual benchmarks, over 29 and 42 languages respectively, reveals that even in low-resource languages, model confidence can increase significantly after instruction-tuning on high-resource language SFT datasets. However, improvements in accuracy are marginal or non-existent, resulting in mis-calibration, highlighting a critical shortcoming of standard SFT for multilingual languages. Furthermore, we observe that the use of label smoothing to be a reasonable method alleviate this concern, again without any need for low-resource SFT data, maintaining better calibration across all languages. Overall, this highlights the importance of multilingual considerations for both training and tuning LLMs in order to improve their reliability and fairness in downstream use.

Large language models (LLMs) often achieve impressive predictive accuracy, yet correctness alone does not imply genuine understanding. True LLM understanding, analogous to human expertise, requires making consistent, well-founded decisions across multiple instances and diverse domains, relying on relevant and domain-grounded decision factors. We introduce Structured Tabular Decision Simulations (STaDS), a suite of expert-like decision settings that evaluate LLMs as if they were professionals undertaking structured decision ``exams''. In this context, understanding is defined as the ability to identify and rely on the correct decision factors, features that determine outcomes within a domain. STaDS jointly assesses understanding through: (i) question and instruction comprehension, (ii) knowledge-based prediction, and (iii) reliance on relevant decision factors. By analyzing 9 frontier LLMs across 15 diverse decision settings, we find that (a) most models struggle to achieve consistently strong accuracy across diverse domains; (b) models can be accurate yet globally unfaithful, and there are frequent mismatches between stated rationales and factors driving predictions. Our findings highlight the need for global-level understanding evaluation protocols and advocate for novel frameworks that go beyond accuracy to enhance LLMs' understanding ability.

--Large language models like GPT -4, Gemini, and Claude have transformed natural language processing (NLP) tasks such as question answering, dialogue generation, summarization, and so forth; yet their susceptibility to hallucination stands as one of the major challenges. Among numerous approaches to overcome this challenge, integration of Knowledge Graphs (KGs) into language models has emerged as a promising solution as it provides structured, reliable, domain-specific, and up-to-date external information to the language models. In this paper, we introduce ALIGNed-LLM, a simple yet effective approach to improve language models' factuality via a lean strategy to infuse KGs into the latent space of language models inspired by LLaV A where visual and textual information is infused. We use embeddings from a pre-trained Knowledge Graph Embedding (KGE) model, such as TransE, and a trainable projection layer to align entity and text embeddings. This alignment enables the language model to distinguish between similar entities improving factual grounding and reducing hallucination. We tested our approach on three popular questions-answering benchmark datasets alongside language models of varying sizes, showing significant improvement. Furthermore, we applied our approach to a real-world financial use case from a large central bank in Europe, which demands high accuracy and precision, demonstrating a substantial improvement of the LLM answers. The emergence of Large Language Models (LLMs) such as GPT -4 [27], Gemini [1], Llama [36], and Claude [2], is producing a performance revolution across natural language processing tasks, such as question answering, dialogue generation, summarization, and many more. LLMs have also confirmed the universality of human language, showing their potential to build domain-specific assistants, trained to follow natural language instructions and accomplish various tasks end to end [22]. The opinions expressed in this paper are personal and should not be attributed to Banca d'Italia. The work was done outside Amazon.

With the wide adoption of large language models (LLMs) in information assistance, it is essential to examine their alignment with human communication styles and values. We situate this study within the context of fact-checking health information, given the critical challenge of rectifying conceptions and building trust. Recent studies have explored the potential of LLM for health communication, but style differences between LLMs and human experts and associated reader perceptions remain under-explored. In this light, our study evaluates the communication styles of LLMs, focusing on how their explanations differ from those of humans in three core components of health communication: information, sender, and receiver. We compiled a dataset of 1498 health misinformation explanations from authoritative fact-checking organizations and generated LLM responses to inaccurate health information. Drawing from health communication theory, we evaluate communication styles across three key dimensions of information linguistic features, sender persuasive strategies, and receiver value alignments. We further assessed human perceptions through a blinded evaluation with 99 participants. Our findings reveal that LLM-generated articles showed significantly lower scores in persuasive strategies, certainty expressions, and alignment with social values and moral foundations. However, human evaluation demonstrated a strong preference for LLM content, with over 60% responses favoring LLM articles for clarity, completeness, and persuasiveness. Our results suggest that LLMs' structured approach to presenting information may be more effective at engaging readers despite scoring lower on traditional measures of quality in fact-checking and health communication.

The significant advancements of Large Language Models (LLMs) in generative tasks have led to a growing body of work exploring LLM-based embedding models. While these models, employing different pooling and attention strategies, have achieved state-of-the-art performance on public embedding benchmarks, questions still arise about what constitutes an effective design for LLM-based embedding models. However, these models are often trained on different datasets, using different LLM base models or training settings. Moreover, evaluations on public embedding benchmarks often fail to report statistical significance, making it difficult to determine which designs truly contribute to final performance. This complicates the process for practitioners seeking optimal training recipes for LLM-based embedding models. In this study, we conduct a large-scale experiment by training a series of LLM-based embedding models using the same training data and base model but differing in their pooling and attention strategies. The results show that there is no one-size-fits-all solution: while bidirectional attention and an additional trainable pooling layer outperform in text similarity and information retrieval tasks, they do not significantly surpass simpler designs like EOS-last token pooling and default causal attention in clustering and classification tasks. Furthermore, we propose a new pooling strategy, Multi-Layers Trainable Pooling, which transforms the outputs of all hidden layers, rather than just the last layer, using a cross-attention network. This method proves to be statistically superior in text similarity and retrieval tasks compared to existing pooling methods. Overall, this paper sheds light on effective training strategies for LLM-based embedding models.

While various approaches have recently been studied for bias identification, little is known about how implicit language that does not explicitly convey a viewpoint affects bias amplification in large language models. To examine the severity of bias toward a view, we evaluated the performance of two downstream tasks where the implicit and explicit knowledge of social groups were used. First, we present a stress test evaluation by using a biased model in edge cases of excessive bias scenarios. Then, we evaluate how LLMs calibrate linguistically in response to both implicit and explicit opinions when they are aligned with conflicting viewpoints. Our findings reveal a discrepancy in LLM performance in identifying implicit and explicit opinions, with a general tendency of bias toward explicit opinions of opposing stances. Moreover, the bias-aligned models generate more cautious responses using uncertainty phrases compared to the unaligned (zero-shot) base models. The direct, incautious responses of the unaligned models suggest a need for further refinement of decisiveness by incorporating uncertainty markers to enhance their reliability, especially on socially nuanced topics with high subjectivity.

Recent advances in large language models (LLMs) have enabled autonomous agents with complex reasoning and task-fulfillment capabilities using a wide range of tools. However, effectively identifying the most relevant tools for a given task becomes a key bottleneck as the toolset size grows, hindering reliable tool utilization. To address this, we introduce Re-Invoke, an unsupervised tool retrieval method designed to scale effectively to large toolsets without training. Specifically, we first generate a diverse set of synthetic queries that comprehensively cover different aspects of the query space associated with each tool document during the tool indexing phase. Second, we leverage LLM's query understanding capabilities to extract key tool-related context and underlying intents from user queries during the inference phase. Finally, we employ a novel multi-view similarity ranking strategy based on intents to pinpoint the most relevant tools for each query. Our evaluation demonstrates that Re-Invoke significantly outperforms state-of-the-art alternatives in both single-tool and multi-tool scenarios, all within a fully unsupervised setting. Notably, on the ToolE datasets, we achieve a 20% relative improvement in nDCG@5 for single-tool retrieval and a 39% improvement for multi-tool retrieval.

In this work, we introduce the Learnable Response Scoring Function (LARS) for Uncertainty Estimation (UE) in generative Large Language Models (LLMs). Current scoring functions for probability-based UE, such as length-normalized scoring and semantic contribution-based weighting, are designed to solve specific aspects of the problem but exhibit limitations, including the inability to handle biased probabilities and under-performance in low-resource languages like Turkish. To address these issues, we propose LARS, a scoring function that leverages supervised data to capture complex dependencies between tokens and probabilities, thereby producing more reliable and calibrated response scores in computing the uncertainty of generations. Our extensive experiments across multiple datasets show that LARS substantially outperforms existing scoring functions considering various probability-based UE methods.

Causal structure discovery from observations can be improved by integrating background knowledge provided by an expert to reduce the hypothesis space. Recently, Large Language Models (LLMs) have begun to be considered as sources of prior information given the low cost of querying them relative to a human expert. In this work, firstly, we propose a set of metrics for assessing LLM judgments for causal graph discovery independently of the downstream algorithm. Secondly, we systematically study a set of prompting designs that allows the model to specify priors about the structure of the causal graph. Finally, we present a general methodology for the integration of LLM priors in graph discovery algorithms, finding that they help improve performance on common-sense benchmarks and especially when used for assessing edge directionality. Our work highlights the potential as well as the shortcomings of the use of LLMs in this problem space.