Large language models are increasingly relied upon as sources of information, but their propensity for generating false or misleading statements with high confidence poses risks for users and society. In this paper, we confront the critical problem of epistemic miscalibration -- where a model's linguistic assertiveness fails to reflect its true internal certainty. We introduce a new human-labeled dataset and a novel method for measuring the linguistic assertiveness of Large Language Models (LLMs) which cuts error rates by over 50% relative to previous benchmarks. Validated across multiple datasets, our method reveals a stark misalignment between how confidently models linguistically present information and their actual accuracy. Further human evaluations confirm the severity of this miscalibration. This evidence underscores the urgent risk of the overstated certainty LLMs hold which may mislead users on a massive scale. Our framework provides a crucial step forward in diagnosing this miscalibration, offering a path towards correcting it and more trustworthy AI across domains. Large Language Models (LLMs) have markedly transformed how humans seek and consume information, becoming integral across diverse fields such as public health (Ali et al., 2023), coding (Zambrano et al., 2023), and education (Whalen & et al., 2023). Despite their growing influence, LLMs are not without shortcomings. One notable issue is the potential for generating responses that, while convincing, may be inaccurate or nonsensical--a long-standing phenomenon often referred to as "hallucinations" (Jo, 2023; Huang et al., 2023; Zhou et al., 2024b). This raises concerns about the reliability and trustworthiness of these models. A critical aspect of trustworthiness in LLMs is epistemic calibration, which represents the alignment between a model's internal confidence in its outputs and the way it expresses that confidence through natural language. Misalignment between internal certainty and external expression can lead to users being misled by overconfident or underconfident statements, posing significant risks in high-stakes domains such as legal advice, medical diagnosis, and misinformation detection. While of great normative concern, how LLMs express linguistic uncertainty has received relatively little attention to date (Sileo & Moens, 2023; Belem et al., 2024). Figures 1 and 5 illustrate the issue of epistemic calibration providing insights into the operation of certainty in the context of human interactions with LLMs. Distinct Roles of Certainty: Internal certainty and linguistic assertiveness have distinct functions within LLM interactions that shape individual beliefs. Human access to LLM certainty: Linguistic assertiveness holds a critical role as the primary form of certainty available to users.

Large Language Models (LLMs) have shown significant advances in text generation but often lack the reliability needed for autonomous deployment in high-stakes domains like healthcare, law, and finance. Existing approaches rely on external knowledge or human oversight, limiting scalability. We introduce a novel framework that repurposes ensemble methods for content validation through model consensus. In tests across 78 complex cases requiring factual accuracy and causal consistency, our framework improved precision from 73.1% to 93.9% with two models (95% CI: 83.5%-97.9%) and to 95.6% with three models (95% CI: 85.2%-98.8%). Statistical analysis indicates strong inter-model agreement ($\kappa$ > 0.76) while preserving sufficient independence to catch errors through disagreement. We outline a clear pathway to further enhance precision with additional validators and refinements. Although the current approach is constrained by multiple-choice format requirements and processing latency, it offers immediate value for enabling reliable autonomous AI systems in critical applications.

Multilingual machine translation models often outperform traditional bilingual models by leveraging translation knowledge transfer. Recent advancements have led to these models supporting hundreds of languages and achieving state-of-the-art results across various translation directions. However, as these models grow larger, their inference operations become increasingly costly. In many use cases, there is no need to support such a wide range of language pairs, as translation is typically needed in only a few selected directions. In this paper, we present CULL-MT, a compression method for machine translation models based on structural layer pruning and selected language directions. Our approach identifies and prunes unimportant layers using a greedy strategy, then mitigates the impact by applying knowledge distillation from the original model along with parameter-efficient fine-tuning. We apply CULL-MT to the NLLB-3.3B and LLaMA3.1-8B-Instruct models. In a multi-way translation scenario (Persian, French, and German to English), we find the NLLB-3.3B model to be robust, allowing 25% of layers to be pruned with only a 0.9 spBLEU drop. However, LLaMA3.1-8B-Instruct is more sensitive, with a 2.0 spBLEU drop after pruning 5 layers.

Auditing Large Language Models (LLMs) to discover their biases and preferences is an emerging challenge in creating Responsible Artificial Intelligence (AI). While various methods have been proposed to elicit the preferences of such models, countermeasures have been taken by LLM trainers, such that LLMs hide, obfuscate or point blank refuse to disclosure their positions on certain subjects. This paper presents PRISM, a flexible, inquiry-based methodology for auditing LLMs - that seeks to illicit such positions indirectly through task-based inquiry prompting rather than direct inquiry of said preferences. To demonstrate the utility of the methodology, we applied PRISM on the Political Compass Test, where we assessed the political leanings of twenty-one LLMs from seven providers. We show LLMs, by default, espouse positions that are economically left and socially liberal (consistent with prior work). We also show the space of positions that these models are willing to espouse - where some models are more constrained and less compliant than others - while others are more neutral and objective. In sum, PRISM can more reliably probe and audit LLMs to understand their preferences, biases and constraints.

Contextual ASR or hotword customization holds substantial practical value. Despite the impressive performance of current end-to-end (E2E) automatic speech recognition (ASR) systems, they often face challenges in accurately recognizing rare words. Typical E2E contextual ASR models commonly feature complex architectures and decoding mechanisms, limited in performance and susceptible to interference from distractor words. With large language model (LLM)-based ASR models emerging as the new mainstream, we propose a CTC-Assisted LLM-Based Contextual ASR model with an efficient filtering algorithm. By using coarse CTC decoding results to filter potential relevant hotwords and incorporating them into LLM prompt input, our model attains WER/B-WER of 1.27%/3.67% and 2.72%/8.02% on the Librispeech test-clean and test-other sets targeting on recognizing rare long-tail words, demonstrating significant improvements compared to the baseline LLM-based ASR model, and substantially surpassing other related work. More remarkably, with the help of the large language model and proposed filtering algorithm, our contextual ASR model still performs well with 2000 biasing words.

As large language models (LLMs) grow and develop, so do their data demands. This is especially true for multilingual LLMs, where the scarcity of high-quality and readily available data online has led to a multitude of synthetic dataset generation approaches. A key technique in this space is machine translation (MT), where high-quality English text is adapted to a target, comparatively low-resource language. This report introduces FineWeb-Edu-Ar, a machine-translated version of the exceedingly popular (deduplicated) FineWeb-Edu dataset from HuggingFace. To the best of our knowledge, FineWeb-Edu-Ar is the largest publicly available machine-translated Arabic dataset out there, with its size of 202B tokens of an Arabic-trained tokenizer.

Large Language Models (LLMs) hold great promise to revolutionize current clinical systems for their superior capacities on medical text processing tasks and medical licensing exams. Meanwhile, traditional ML models such as SVM and XGBoost have still been mainly adopted in clinical prediction tasks. An emerging question is Can LLMs beat traditional ML models in clinical prediction? Thus, we build a new benchmark ClinicalBench to comprehensively study the clinical predictive modeling capacities of both general-purpose and medical LLMs, and compare them with traditional ML models. ClinicalBench embraces three common clinical prediction tasks, two databases, 14 general-purpose LLMs, 8 medical LLMs, and 11 traditional ML models. Through extensive empirical investigation, we discover that both general-purpose and medical LLMs, even with different model scales, diverse prompting or fine-tuning strategies, still cannot beat traditional ML models in clinical prediction yet, shedding light on their potential deficiency in clinical reasoning and decision-making. We call for caution when practitioners adopt LLMs in clinical applications. ClinicalBench can be utilized to bridge the gap between LLMs' development for healthcare and real-world clinical practice.

We present a method to compress the final linear layer of language models, reducing memory usage by up to 3.4x without significant performance loss. By grouping tokens based on Byte Pair Encoding (BPE) merges, we prevent materialisation of the memory-intensive logits tensor. Evaluations on the TinyStories dataset show that our method performs on par with GPT-Neo and GPT2 while significantly improving throughput by up to 3x, making it suitable for low-compute environments.

Large Language Models (LLMs) have achieved significant success in various natural language processing tasks, but the role of wireless networks in supporting LLMs has not been thoroughly explored. In this paper, we propose a wireless distributed Mixture of Experts (WDMoE) architecture to enable collaborative deployment of LLMs across edge servers at the base station (BS) and mobile devices in wireless networks. Specifically, we decompose the MoE layer in LLMs by placing the gating network and the preceding neural network layer at BS, while distributing the expert networks among the devices. This deployment leverages the parallel inference capabilities of expert networks on mobile devices, effectively utilizing the limited computing and caching resources of these devices. Accordingly, we develop a performance metric for WDMoE-based LLMs, which accounts for both model capability and latency. To minimize the latency while maintaining accuracy, we jointly optimize expert selection and bandwidth allocation based on the performance metric. Moreover, we build a hardware testbed using NVIDIA Jetson kits to validate the effectiveness of WDMoE. Both theoretical simulations and practical hardware experiments demonstrate that the proposed method can significantly reduce the latency without compromising LLM performance.

Since the COVID-19 pandemic, clinicians have seen a large and sustained influx in patient portal messages, significantly contributing to clinician burnout. To the best of our knowledge, there are no large-scale public patient portal messages corpora researchers can use to build tools to optimize clinician portal workflows. Informed by our ongoing work with a regional hospital, this study introduces an LLM-powered framework for configurable and realistic patient portal message generation. Our approach leverages few-shot grounded text generation, requiring only a small number of de-identified patient portal messages to help LLMs better match the true style and tone of real data. Clinical experts in our team deem this framework as HIPAA-friendly, unlike existing privacy-preserving approaches to synthetic text generation which cannot guarantee all sensitive attributes will be protected. Through extensive quantitative and human evaluation, we show that our framework produces data of higher quality than comparable generation methods as well as all related datasets. We believe this work provides a path forward for (i) the release of large-scale synthetic patient message datasets that are stylistically similar to ground-truth samples and (ii) HIPAA-friendly data generation which requires minimal human de-identification efforts.