Wikipedia is the largest open knowledge corpus, widely used worldwide and serving as a key resource for training large language models (LLMs) and retrieval-augmented generation (RAG) systems. Ensuring its accuracy is therefore critical. But how accurate is Wikipedia, and how can we improve it? We focus on inconsistencies, a specific type of factual inaccuracy, and introduce the task of corpus-level inconsistency detection. We present CLAIRE, an agentic system that combines LLM reasoning with retrieval to surface potentially inconsistent claims along with contextual evidence for human review. In a user study with experienced Wikipedia editors, 87.5% reported higher confidence when using CLAIRE, and participants identified 64.7% more inconsistencies in the same amount of time. Combining CLAIRE with human annotation, we contribute WIKICOLLIDE, the first benchmark of real Wikipedia inconsistencies. Using random sampling with CLAIRE-assisted analysis, we find that at least 3.3% of English Wikipedia facts contradict another fact, with inconsistencies propagating into 7.3% of FEVEROUS and 4.0% of AmbigQA examples. Benchmarking strong baselines on this dataset reveals substantial headroom: the best fully automated system achieves an AUROC of only 75.1%. Our results show that contradictions are a measurable component of Wikipedia and that LLM-based systems like CLAIRE can provide a practical tool to help editors improve knowledge consistency at scale.

Ultrasound tongue imaging (UTI) is a non-invasive and cost-effective tool for studying speech articulation, motor control, and related disorders. However, real-time tongue contour segmentation remains challenging due to low signal-to-noise ratios, imaging variability, and computational demands. We propose UltraUNet, a lightweight encoder-decoder architecture optimized for real-time segmentation of tongue contours in ultrasound images. UltraUNet incorporates domain-specific innovations such as lightweight Squeeze-and-Excitation blocks, Group Normalization for small-batch stability, and summation-based skip connections to reduce memory and computational overhead. It achieves 250 frames per second and integrates ultrasound-specific augmentations like denoising and blur simulation. Evaluations on 8 datasets demonstrate high accuracy and robustness, with single-dataset Dice = 0.855 and MSD = 0.993px, and cross-dataset Dice averaging 0.734 and 0.761. UltraUNet provides a fast, accurate solution for speech research, clinical diagnostics, and analysis of speech motor disorders.

Large language models (LLMs) excel at general mathematical reasoning but fail catastrophically on specialized technical mathematics. In wireless communications, where problems require precise manipulation of information-theoretic bounds, optimization constraints, and signal processing formulations, even state-of-the-art models struggle to achieve competent performance. We present WirelessMathLM, demonstrating that compact models (0.5B-7B parameters) can match or exceed much larger models through domain-specific reinforcement learning with verifiable rewards. Our key insight is that wireless mathematics problems possess a unique property--verifiable correctness--that enables effective reinforcement learning without human feedback. We construct WirelessMathBench-XL, a comprehensive benchmark of 4,027 problems from 970 papers. Using Group Relative Policy Optimization (GRPO) with binary verification rewards, we train models directly from base checkpoints without supervised warm-start. Our 7B model achieves 39.5% accuracy on WirelessMathBench-XL, approaching GPT-4o (40.4%) while using about 100 times fewer parameters than DeepSeek-R1 (671B, 57.4%). Remarkably, GRPO training nearly doubles performance across all model scales (0.5B +11%, 3B +103%, 7B +81%), with positive transfer to general mathematics benchmarks--our models gain +8.4 points on average across MATH, Minerva-Math, OlympiadBench, AMC, and AIME without any training on these tasks.

Language Models, when generating prepositional phrases, must often decide for whether their complements functions as an instrumental adjunct (describing the verb adverbially) or an attributive modifier (enriching the noun adjectivally), yet the internal mechanisms that resolve this split decision remain poorly understood. In this study, we conduct a targeted investigation into Gemma-2 to uncover and control the generation of prepositional complements. We assemble a prompt suite containing with-headed prepositional phrases whose contexts equally accommodate either an instrumental or attributive continuation, revealing a strong preference for an instrumental reading at a ratio of 3:4. To pinpoint individual attention heads that favor instrumental over attributive complements, we project activations into the vocabulary space. By scaling the value vector of a single attention head, we can shift the distribution of functional roles of complements, attenuating instruments to 33% while elevating attributes to 36%.

Recent reasoning LLMs (RLMs), especially those trained with verifier-based reinforcement learning, often perform worse with few-shot CoT than with direct answering. We revisit this paradox using high-quality reasoning traces from DeepSeek-R1 as demonstrations and find that adding more exemplars consistently degrades accuracy, even when demonstrations are optimal. A detailed analysis reveals two mechanisms behind this decline: (i) semantic misguidance, where high textual similarity leads the model to treat the target as the same as the exemplar and to copy intermediate steps verbatim; and (ii) strategy transfer failure, where the model struggles to extract useful reasoning strategies and apply them to target questions. Guided by these, we introduce Insight-to-Solve (I2S), a sequential test-time procedure that turns demonstrations into explicit, reusable insights and derives a target-specific reasoning trace; optionally, the reasoning is self-refined for coherence and correctness (I2S+). Extensive experiments on diverse benchmarks show that I2S and I2S+ consistently outperform both direct answering and test-time scaling baselines across open- and closed-source models. Even for GPT models, our method helps: on AIME'25, GPT-4.1 rises by +14.0%, and o1-mini improves by +2.7% on AIME and +1.7% on GPQA, indicating that in-context demonstrations can be harnessed effectively via insight-refine-solve framework.

Large Language Models (LLMs) have achieved impressive performance across diverse tasks but continue to struggle with learning transitive relations, a cornerstone for complex planning. To address this issue, we investigate the Multi-Token Prediction (MTP) paradigm and its impact to transitive relation learning. We theoretically analyze the MTP paradigm using a Transformer architecture composed of a shared output head and a transfer layer. Our analysis reveals that the transfer layer gradually learns the multi-step adjacency information, which in turn enables the backbone model to capture unobserved transitive reachability relations beyond those directly present in the training data, albeit with some inevitable noise in adjacency estimation. Building on this foundation, we propose two strategies to enhance the transfer layer and overall learning quality: Next-Token Injection (NTI) and a Transformer-based transfer layer. Our experiments on both synthetic graphs and the Blocksworld planning benchmark validate our theoretical findings and demonstrate that the improvements significantly enhance the model's path-planning capability. These findings deepen our understanding of how Transformers with MTP learn in complex planning tasks, and provide practical strategies to overcome the transitivity bottleneck, paving the way toward structurally aware and general-purpose planning models.

Test-time entropy minimization helps adapt a model to novel environments and incentivize its reasoning capability, unleashing the model's potential during inference by allowing it to evolve and improve in real-time using its own predictions, achieving promising performance. However, pure entropy minimization can favor non-generalizable shortcuts, such as inflating the logit norm and driving all predictions to a dominant class to reduce entropy, risking collapsed solutions (e.g., constant one-hot outputs) that trivially minimize the objective without meaningful learning. In this paper, we introduce ZeroSiam, an efficient asymmetric Siamese architecture tailored for test-time entropy minimization. ZeroSiam prevents collapse through asymmetric divergence alignment, which is efficiently achieved by a learnable predictor and a stop-gradient operator before the classifier. We provide empirical and theoretical evidence that ZeroSiam not only prevents collapse solutions, but also absorbs and regularizes biased learning signals, enhancing performance even when no collapse occurs. Despite its simplicity, extensive results show that ZeroSiam performs more stably over prior methods using negligible overhead, demonstrating efficacy on both vision adaptation and large language model reasoning tasks across challenging test scenarios and diverse models, including tiny models that are particularly collapse-prone.

The transformer architecture introduced by [V aswani et al., 2017] has demonstrated capabilities across a wide range of tasks [Liu et al., 2018, Devlin et al., 2019, Rad-ford et al., 2019, Touvron et al., 2023, OpenAI, 2023], showing particularly significant progress in logical reasoning. These models can not only solve complex mathematical problems Davies et al. [2021] but have also reached performance levels comparable to top human contestants in the International Mathematical Olympiad (IMO) [Trinh et al., 2024]. The reasoning capabilities of large language models are fundamentally shaped by the thinking strategies they employ.

Abstract--With the emergence of Web 2.0 and microservices architecture, the number of Web APIs has increased dramatically, further intensifying the demand for efficient Web API recommendation. Existing solutions typically fall into two categories: recommendation-type methods, which treat each API as a label for classification, and match-type methods, which focus on matching mashups through API retrieval. However, three critical challenges persist: 1) the semantic ambiguities in comparing API and mashup descriptions, 2) the lack of detailed comparisons between the individual API and the mashup in recommendation-type methods, and 3) time inefficiencies for API retrieval in match-type methods. T o address these challenges, we propose W ARBERT, a hierarchical BERT -based model for Web API recommendation. W ARBERT leverages dual-component feature fusion and attention comparison to extract precise semantic representations of API and mashup descriptions. W ARBERT consists of two main components: W ARBERT(R) for Recommendation and W ARBERT(M) for Matching. Specifically, W AR-BERT(R) serves as an initial filter, narrowing down the candidate APIs, while W ARBERT(M) refines the matching process by calculating the similarity between candidate APIs and mashup. The final likelihood of a mashup being matched with an API is determined by combining the predictions from W ARBERT(R) and W ARBERT(M). Additionally, W ARBERT(R) incorporates an auxiliary task of mashup category judgment, which enhances its effectiveness in candidate selection. Experimental results on the ProgrammableWeb dataset demonstrate that W ARBERT outperforms most existing solutions and achieves improvements of up to 11.7% compared to the model MTFM (Multi-T ask Fusion Model), delivering significant enhancements in accuracy and efficiency. ITH the emergence of Web 2.0 and microservice architecture, the number of APIs has increased dramatically [1]. Since 2022, there have been more than 24,000 APIs in ProgrammableWeb [2]. The benefits of Web APIs have led to the emergence of a novel method for developing applications, known as Mashup [3]. Mashup enables developers to integrate existing Web API resources to meet complex requirements without starting from scratch [4], [5].

Accurate counting of vehicle axles is essential for traffic control, toll collection, and infrastructure development. We present an end-to-end, video-based pipeline for axle counting that tackles limitations of previous works in dense environments. Our system leverages a combination of YOLO-OBB to detect and categorize vehicles, and YOLO to detect tires. Detected tires are intelligently associated to their respective parent vehicles, enabling accurate axle prediction even in complex scenarios. However, there are a few challenges in detection when it comes to scenarios with longer and occluded vehicles. We mitigate vehicular occlusions and partial detections for longer vehicles by proposing a novel TRAX (Tire and Axle Tracking) Algorithm to successfully track axle-related features between frames. Our method stands out by significantly reducing false positives and improving the accuracy of axle-counting for long vehicles, demonstrating strong robustness in real-world traffic videos. This work represents a significant step toward scalable, AI-driven axle counting systems, paving the way for machine vision to replace legacy roadside infrastructure.