Real-world processes often generate data that are a mix of categorical and numeric values that are recorded at irregular and informative intervals. Discrete token-based approaches are limited in numeric representation capacity while methods like neural ordinary differential equations are not well suited for categorical data or informative sampling and require augmentation to handle certain classes of trajectories. Here, we present multivariateGPT, a single architecture for modeling sequences of mixed categorical (including tokenized text) and numeric data. This is accomplished with an autoregressive sequence decomposition, embedding scheme, and loss function that extend the next token prediction task to likelihood estimation of the joint distribution of next token class and value. We demonstrate how this approach can efficiently learn to generalize patterns in simple physical systems and model complex time series including electrocardiograms and multivariate electronic health record data. This work extends the utility of transformer based models to additional classes of data.

--Encrypted traffic classification is highly challenging in network security due to the need for extracting robust features from content-agnostic traffic data. Existing approaches face critical issues: (i) Distribution drift, caused by reliance on the closed-world assumption, limits adaptability to real-world, shifting patterns; (ii) Dependence on labeled data restricts applicability where such data is scarce or unavailable. Large language models (LLMs) have demonstrated remarkable potential in offering generalizable solutions across a wide range of tasks, achieving notable success in various specialized fields. However, their effectiveness in traffic analysis remains constrained by challenges in adapting to the unique requirements of the traffic domain. In this paper, we introduce a novel traffic representation model named Encrypted Traffic Out-of-Distribution Instruction T uning with LLM (ET ooL), which integrates LLMs with knowledge of traffic structures through a self-supervised instruction tuning paradigm. This framework establishes connections between textual information and traffic interactions. ET ooL demonstrates more robust classification performance and superior generalization in both supervised and zero-shot traffic classification tasks. Additionally, we construct NETD, a traffic dataset designed to support dynamic distributional shifts, and use it to validate ET ooL's effectiveness under varying distributional conditions. Furthermore, we evaluate the efficiency gains achieved through ET ooL's instruction tuning approach. Received 22 October 2024; revised 30 April 2025; accepted 20 May 2025. This work is supported by The National Key Research and Development Program of China No. 2024YFF1401300. Gang Xiong, Gaopeng Gou, Wenqi Dong, Jing Y u, Zhen Li and Wei Xia are with the Institute of Information Engineering, Chinese Academy of Sciences, Beijing 100190, China, and also with the School of Cyber Security, University of Chinese Academy of Sciences, Beijing 100085, China. In recent years, gradual full encryption of traffic has become a reality, explicit fingerprinting has been gradually failing. Different technical approaches have been proposed to address the needs of encrypted traffic analysis, including: ( i) Statistical feature-based approaches [15], [39] extract statistical features and combine them with classical machine learning algorithms to cope with traffic without plaintext; ( ii) Raw feature-based approaches [8], [22] on the other hand selects raw traffic features and captures complicated patterns based on deep learning algorithms; and ( iii) Raw datagram-based approaches [18]- [20] utilize deep neural networks to learn implicit correlations between datagram bytes.

In this paper, we propose a novel few-shot optimization with HED-LM (Hybrid Euclidean Distance with Large Language Models) to improve example selection for sensor-based classification tasks. While few-shot prompting enables efficient inference with limited labeled data, its performance largely depends on the quality of selected examples. HED-LM addresses this challenge through a hybrid selection pipeline that filters candidate examples based on Euclidean distance and re-ranks them using contextual relevance scored by large language models (LLMs). To validate its effectiveness, we apply HED-LM to a fatigue detection task using accelerometer data characterized by overlapping patterns and high inter-subject variability. Unlike simpler tasks such as activity recognition, fatigue detection demands more nuanced example selection due to subtle differences in physiological signals. Our experiments show that HED-LM achieves a mean macro F1-score of 69.13$\pm$10.71%, outperforming both random selection (59.30$\pm$10.13%) and distance-only filtering (67.61$\pm$11.39%). These represent relative improvements of 16.6% and 2.3%, respectively. The results confirm that combining numerical similarity with contextual relevance improves the robustness of few-shot prompting. Overall, HED-LM offers a practical solution to improve performance in real-world sensor-based learning tasks and shows potential for broader applications in healthcare monitoring, human activity recognition, and industrial safety scenarios.

Advancements in Large Language Models (LLMs) drive interest in scientific applications, necessitating specialized benchmarks such as Earth science. Existing benchmarks either present a general science focus devoid of Earth science specificity or cover isolated subdomains, lacking holistic evaluation. Furthermore, current benchmarks typically neglect the assessment of LLMs' capabilities in open-ended scientific exploration. In this paper, we present a comprehensive and professional benchmark for the Earth sciences, designed to evaluate the capabilities of LLMs in scientific exploration within this domain, spanning from fundamental to advanced levels. Leveraging a corpus of 100,000 research papers, we first construct two Question Answering (QA) datasets: Earth-Iron, which offers extensive question coverage for broad assessment, and Earth-Silver, which features a higher level of difficulty to evaluate professional depth. These datasets encompass five Earth spheres, 114 disciplines, and 11 task categories, assessing foundational knowledge crucial for scientific exploration. Most notably, we introduce Earth-Gold with new metrics, a dataset comprising open-ended multi-turn dialogues specifically designed to evaluate the advanced capabilities of LLMs in scientific exploration, including methodology induction, limitation analysis, and concept proposal. Extensive experiments reveal limitations in 11 leading LLMs across different domains and tasks, highlighting considerable room for improvement in their scientific exploration capabilities. The benchmark is available on https://huggingface.co/ai-earth .

Factual hallucinations are a major challenge for Large Language Models (LLMs). They undermine reliability and user trust by generating inaccurate or fabricated content. Recent studies suggest that when generating false statements, the internal states of LLMs encode information about truthfulness. However, these studies often rely on synthetic datasets that lack realism, which limits generalization when evaluating the factual accuracy of text generated by the model itself. In this paper, we challenge the findings of previous work by investigating truthfulness encoding capabilities, leading to the generation of a more realistic and challenging dataset. Specifically, we extend previous work by introducing: (1) a strategy for sampling plausible true-false factoid sentences from tabular data and (2) a procedure for generating realistic, LLM-dependent true-false datasets from Question Answering collections. Our analysis of two open-source LLMs reveals that while the findings from previous studies are partially validated, generalization to LLM-generated datasets remains challenging. This study lays the groundwork for future research on factuality in LLMs and offers practical guidelines for more effective evaluation.

Recently, reasoning-based MLLMs have achieved a degree of success in generating long-form textual reasoning chains. However, they still struggle with complex tasks that necessitate dynamic and iterative focusing on and revisiting of visual regions to achieve precise grounding of textual reasoning in visual evidence. We introduce \textbf{VLM-R$^3$} (\textbf{V}isual \textbf{L}anguage \textbf{M}odel with \textbf{R}egion \textbf{R}ecognition and \textbf{R}easoning), a framework that equips an MLLM with the ability to (i) decide \emph{when} additional visual evidence is needed, (ii) determine \emph{where} to ground within the image, and (iii) seamlessly weave the relevant sub-image content back into an interleaved chain-of-thought. The core of our method is \textbf{Region-Conditioned Reinforcement Policy Optimization (R-GRPO)}, a training paradigm that rewards the model for selecting informative regions, formulating appropriate transformations (e.g.\ crop, zoom), and integrating the resulting visual context into subsequent reasoning steps. To bootstrap this policy, we compile a modest but carefully curated Visuo-Lingual Interleaved Rationale (VLIR) corpus that provides step-level supervision on region selection and textual justification. Extensive experiments on MathVista, ScienceQA, and other benchmarks show that VLM-R$^3$ sets a new state of the art in zero-shot and few-shot settings, with the largest gains appearing on questions demanding subtle spatial reasoning or fine-grained visual cue extraction.

Prompt engineering for large language models is challenging, as even small prompt perturbations or model changes can significantly impact the generated output texts. Existing evaluation methods of LLM outputs, either automated metrics or human evaluation, have limitations, such as providing limited insights or being labor-intensive. We propose Spotlight, a new approach that combines both automation and human analysis. Based on data mining techniques, we automatically distinguish between random (decoding) variations and systematic differences in language model outputs. This process provides token patterns that describe the systematic differences and guide the user in manually analyzing the effects of their prompts and changes in models efficiently. We create three benchmarks to quantitatively test the reliability of token pattern extraction methods and demonstrate that our approach provides new insights into established prompt data. From a human-centric perspective, through demonstration studies and a user study, we show that our token pattern approach helps users understand the systematic differences of language model outputs. We are further able to discover relevant differences caused by prompt and model changes (e.g. related to gender or culture), thus supporting the prompt engineering process and human-centric model behavior research.

We study the robustness of Transformer language models under semantic out-of-distribution (OOD) shifts, where training and test data lie in disjoint latent spaces. Using Wasserstein-1 distance and Gevrey-class smoothness, we derive sub-exponential upper bounds on prediction error. Our theoretical framework explains how smoothness governs generalization under distributional drift. We validate these findings through controlled experiments on arithmetic and Chain-of-Thought tasks with latent permutations and scalings. Results show empirical degradation aligns with our bounds, highlighting the geometric and functional principles underlying OOD generalization in Transformers.

Large language models (LLMs) exhibit excellent performance in natural language processing (NLP), but remain highly sensitive to the quality of input queries, especially when these queries contain misleading or inaccurate information. Existing methods focus on correcting the output, but they often overlook the potential of improving the ability of LLMs to detect and correct misleading content in the input itself. In this paper, we propose a novel three-stage fine-tuning method that enhances the ability of LLMs to detect and correct misleading information in the input, further improving response accuracy and reducing hallucinations. Specifically, the three stages include (1) training LLMs to identify misleading information, (2) training LLMs to correct the misleading information using built-in or external knowledge, and (3) training LLMs to generate accurate answers based on the corrected queries. To evaluate our method, we conducted experiments on three datasets for the hallucination detection task and the question answering~(QA) task, as well as two datasets containing misleading information that we constructed. The experimental results demonstrate that our method significantly improves the accuracy and factuality of LLM responses, while also enhancing the ability to detect hallucinations and reducing the generation of hallucinations in the output, particularly when the query contains misleading information.

Reasoning-enabled large language models (LLMs) excel in logical tasks, yet their utility for evaluating natural language generation remains unexplored. This study systematically compares reasoning LLMs with non-reasoning counterparts across machine translation and text summarization evaluation tasks. We evaluate eight models spanning state-of-the-art reasoning models (DeepSeek-R1, OpenAI o3), their distilled variants (8B-70B parameters), and equivalent non-reasoning LLMs. Experiments on WMT23 and SummEval benchmarks reveal architecture and task-dependent benefits: OpenAI o3-mini models show improved performance with increased reasoning on MT, while DeepSeek-R1 and generally underperforms compared to its non-reasoning variant except in summarization consistency evaluation. Correlation analysis demonstrates that reasoning token usage correlates with evaluation quality only in specific models, while almost all models generally allocate more reasoning tokens when identifying more quality issues. Distillation maintains reasonable performance up to 32B parameter models but degrades substantially at 8B scale. This work provides the first assessment of reasoning LLMs for NLG evaluation and comparison to non-reasoning models. We share our code to facilitate further research: https://github.com/NL2G/reasoning-eval.