LangTrend

The Korean writing system, \textit{Hangeul}, has a unique character representation rigidly following the invention principles recorded in \textit{Hunminjeongeum}.\footnote{\textit{Hunminjeongeum} is a book published in 1446 that describes the principles of invention and usage of \textit{Hangeul}, devised by King Sejong \cite{Hunminjeongeum_Guide}.} However, existing pre-trained language models (PLMs) for Korean have overlooked these principles. In this paper, we introduce a novel framework for Korean PLMs called KOMBO, which firstly brings the invention principles of \textit{Hangeul} to represent character. Our proposed method, KOMBO, exhibits notable experimental proficiency across diverse NLP tasks. In particular, our method outperforms the state-of-the-art Korean PLM by an average of 2.11\% in five Korean natural language understanding tasks. Furthermore, extensive experiments demonstrate that our proposed method is suitable for comprehending the linguistic features of the Korean language. Consequently, we shed light on the superiority of using subcharacters over the typical subword-based approach for Korean PLMs. Our code is available at: [https://github.com/SungHo3268/KOMBO](https://github.com/SungHo3268/KOMBO).

In autoregressive large language models (LLMs), temporal straightening offers an account of how the next-token prediction objective shapes representations. Models learn to progressively straighten the representational trajectory of input sequences across layers, potentially facilitating next-token prediction via linear extrapolation. However, a direct link between this trajectory and token-level behavior has been missing. We provide such a link by relating contextual curvature-a geometric measure of how sharply the representational trajectory bends over recent context-to next-token entropy. Across two models (GPT-2 XL and Pythia-2.8B), contextual curvature is correlated with entropy, and this relationship emerges during training. Perturbation experiments reveal selective dependence: manipulating curvature through trajectory-aligned interventions reliably modulates entropy, while geometrically misaligned perturbations have no effect. Finally, regularizing representations to be straighter during training modestly reduces token-level entropy without degrading validation loss. These results identify trajectory curvature as a task-aligned representational feature that influences behavioral uncertainty in LLMs.

Static benchmarks measure what AI agents can do at a fixed point in time but not how they are adopted, maintained, or experienced in deployment. We introduce AgentPulse, a continuous evaluation framework scoring 50 agents across 10 workload categories along four factors (Benchmark Performance, Adoption Signals, Community Sentiment, and Ecosystem Health) aggregated from 18 real-time signals across GitHub, package registries, IDE marketplaces, social platforms, and benchmark leaderboards. Three analyses ground the framework. The four factors capture largely complementary information (n=50; $ρ_{\max}=0.61$ for Adoption-Ecosystem, all others $|ρ| \leq 0.37$). A circularity-controlled test (n=35) shows the Benchmark+Sentiment sub-composite, which contains no GitHub-derived signals, predicts external adoption proxies it does not aggregate: GitHub stars ($ρ_s=0.52$, $p<0.01$) and Stack Overflow question volume ($ρ_s=0.49$, $p<0.01$), with VS Code installs ($ρ_s=0.44$, $p<0.05$) reported as illustrative given that only 11 of 35 agents have non-zero installs. On the n=11 subset with published SWE-bench scores, composite and benchmark-only rankings are nearly uncorrelated ($ρ_s=0.25$; 9 of 11 agents shift by at least 2 ranks), driven by a strong negative Adoption-Capability correlation among closed-source high-capability agents within this subset. This is precisely why we rest the framework's validity claim on the broader n=35 test rather than the SWE-bench overlap. AgentPulse surfaces deployment signal absent from benchmarks; it is a methodology, not a ground-truth ranking. The framework, all collected signals, scoring outputs, and evaluation harness are released under CC BY 4.0.

Transformer-based table retrieval systems flatten structured tables into token sequences, making retrieval sensitive to the choice of serialization even when table semantics remain unchanged. We show that semantically equivalent serializations, such as $\texttt{csv}$, $\texttt{tsv}$, $\texttt{html}$, $\texttt{markdown}$, and $\texttt{ddl}$, can produce substantially different embeddings and retrieval results across multiple benchmarks and retriever families. To address this instability, we treat serialization embedding as noisy views of a shared semantic signal and use its centroid as a canonical target representation. We show that centroid averaging suppresses format-specific variation and can recover the semantic content common to different serializations when format-induced shifts differ across tables. Empirically, centroid representations outrank individual formats in aggregate pairwise comparisons across $\texttt{MPNet}$, $\texttt{BGE-M3}$, $\texttt{ReasonIR}$, and $\texttt{SPLADE}$. We further introduce a lightweight residual bottleneck adapter on top of a frozen encoder that maps single-serialization embeddings towards centroid targets while preserving variance and enforcing covariance regularization. The adapter improves robustness for several dense retrievers, though gains are model-dependent and weaker for sparse lexical retrieval. These results identify serialization sensitivity as a major source of retrieval variance and show the promise of post hoc geometric correction for serialization-invariant table retrieval.

Small instruct-tuned LLMs produce degenerate verbal confidence under minimal elicitation: ceiling rates above 95%, near-chance Type-2 AUROC, and Invalid validity profiles. We test whether confidence-conditioned supervised fine-tuning (CSFT) with self-consistency-derived targets can close the gap between internal information and verbal readout. A pre-registered Phase 0 protocol on Gemma 3 4B-it with a modal filter restricting training to items with correct modal answers produced a negative result: AUROC2 dropped from 0.554 to 0.509 due to label-entropy collapse in the training targets. An exploratory rescue removed the filter, training on all 2,000 calibration items. This produced a binary verbal correctness discriminator with AUROC2 = 0.774 on held-out TriviaQA, compressing a 10-sample self-consistency signal (AUROC2 = 0.999) into a single-pass readout exceeding logit entropy (0.701). The shuffled-target control showed no improvement (0.501). On MMLU, accuracy improved from 54.2% to 77.4% with the shuffled model at baseline (56.1%), supporting a target-dependent interpretation. The result is exploratory, binary rather than continuously calibrated, and observed at a single scale. It identifies two design lessons: confidence training requires label entropy, and correct targets regularise output format.

Large Language Models (LLMs) reveal inherent and distinctive personas through dialogue. However, most existing persona discovery approaches rely on surface-level lexical or stylistic cues, treating dialogue as a flat sequence of tokens and failing to capture the deeper discourse-level structures that sustain persona consistency. To address this limitation, we propose a novel analytical framework that interprets LLM dialogue through bridging inference – implicit conceptual relations that connect utterances via shared world knowledge and discourse coherence. By modeling these relations as structured knowledge graphs, our approach captures latent semantic links that govern how LLMs organize meaning across turns, enabling persona discovery at the level of discourse coherence rather than surface realizations. Experimental results across multiple reasoning backbones and target LLMs, ranging from small-scale models to 80B-parameter systems, demonstrate that bridging-inference graphs yield significantly stronger semantic coherence and more stable persona identification than frequency or style-based baselines. These results show that persona traits are consistently encoded in the structural organization of discourse rather than isolated lexical patterns. This work presents a systematic framework for probing, extracting, and visualizing latent LLM personas through the lens of Cognitive Discourse Theory, bridging computational linguistics, cognitive semantics, and persona reasoning in large language models. Codes are available at https://github.com/JiSoo-Yang/Persona_Bridging.git

Fine-tuned autoregressive models for graph-to-sequence generation (G2S) often struggle with factual grounding and edit sensitivity. To tackle these issues, we propose a non-autoregressive diffusion framework that generates text by iterative refinement conditioned on an input graph, named as Diffusion Language Model for Graphs (DLM4G). By aligning graph components (entities/relations) with their corresponding sequence tokens, DLM4G employs an adaptive noising strategy. The proposed strategy uses per-token denoising error as a signal to adaptively modulate noise on entity and relation tokens, improving preservation of graph structure and enabling localized updates under graph edits. Evaluated on three datasets, DLM4G consistently outperforms competitive G2S diffusion baselines trained on identical splits across both surface-form and embedding-based metrics. DLM4G further exceeds fine-tuned autoregressive baselines up to 12x larger (e.g., T5-Large) and is competitive with zero-shot LLM transfer baselines up to 127x larger. Relative to the strongest fine-tuned PLM baseline, DLM4G improves factual grounding (FGT@0.5) by +5.16% and edit sensitivity (ESR) by +7.9%; compared to the best diffusion baseline, it yields gains of +3.75% in FGT@0.5 and +23.6% in ESR. We additionally demonstrate applicability beyond textual graphs through experiments on molecule captioning, indicating the method's generality for scientific G2S generation.

Curriculum learning helps language models tackle complex reasoning by gradually increasing task difficulty. However, it often fails to generate consistent step-by-step reasoning, especially in multilingual and low-resource settings where cross-lingual transfer from English to Indian languages remains limited. We propose IRIS: Interleaved Reinforcement with Incremental Staged Curriculum, a two-axis framework that combines Supervised Fine-Tuning on progressively harder problems (vertical axis) with Reverse Curriculum Reinforcement Learning to reduce reliance on step-by-step guidance (horizontal axis). We design a composite reward combining correctness, step-wise alignment, continuity, and numeric incentives, optimized via Group Relative Policy Optimization (GRPO). We release CL-Math, a dataset of 29k problems with step-level annotations in English, Hindi, and Marathi. Across standard benchmarks and curated multilingual test sets, IRIS consistently improves performance, with strong results on math reasoning tasks and substantial gains in low-resource and bilingual settings, alongside modest improvements in high-resource languages.

Large language models have demonstrated strong reasoning capabilities in general knowledge question answering. However, their ability to handle temporal information remains limited. To address this limitation, existing approaches often involve external tools or manual verification and are tailored to specific scenarios, leading to poor generalizability. Moreover, these methods apply a fixed pipeline to all questions, overlooking the fact that different types of temporal questions require distinct reasoning strategies, which leads to unnecessary processing for simple cases and inadequate reasoning for complex ones. To this end, we propose AdapTime, an adaptive temporal reasoning method that dynamically executes reasoning steps based on the input context. Specifically, it involves three temporal reasoning actions: reformulate, rewrite and review, with an LLM planner guiding the reasoning process. AdapTime integrates seamlessly with state-of-the-art LLMs and significantly enhances their temporal reasoning capabilities without relying on external support. Extensive experiments demonstrate the effectiveness of our approach.

Detecting hate speech in memes is challenging due to their multimodal nature and subtle, culturally grounded cues such as sarcasm and context. While recent vision-language models (VLMs) enable joint reasoning over text and images, end-to-end prompting can be brittle, as a single prediction must resolve target, stance, implicitness, and irony. These challenges are amplified in multilingual settings. We propose a prompted weak supervision (PWS) approach that decomposes meme understanding into targeted, question-based labeling functions with constrained answer options for homophobia and transphobia detection in the LT-EDI 2026 shared task. Using a quantized Qwen3-VLM to extract features by answering targeted questions, our method outperforms direct VLM classification, with substantial gains for Chinese and Hindi, ranking 1st in English, 2nd in Chinese, and 3rd in Hindi. Iterative refinement via error-driven LF expansion and feature pruning reduces redundancy and improves generalization. Our results highlight the effectiveness of prompted weak supervision for multilingual multimodal hate speech detection.

Frontier image generation has moved from artistic synthesis toward synthetic visual evidence. Systems such as GPT Image 2, Nano Banana Pro, Nano Banana 2, Grok Imagine, Qwen Image 2.0 Pro, and Seedream 5.0 Lite combine photorealistic rendering, readable typography, reference consistency, editing control, and in several cases reasoning or search-grounded image construction. These capabilities create large benefits for design, education, accessibility, and communication, yet they also weaken one of society's most common trust shortcuts: the belief that a plausible picture is a reliable record. This paper provides a source-grounded technical and policy analysis of synthetic visual risk. We first summarize the public capabilities of recent image models, then analyze public incidents involving fake crisis images, celebrity and public-figure imagery, medical scans, forged-looking documents, synthetic screenshots, phishing assets, and market-moving rumors. We introduce a capability-weighted risk framework that links model affordances to real-world harm in finance, medicine, news, law, emergency response, identity verification, and civic discourse. Our findings show that risk is driven less by photorealism alone than by the convergence of realism, legible text, identity persistence, fast iteration, and distribution context. We argue for layered control: model-side restrictions, cryptographic provenance, visible labeling, platform friction, sector-grade verification, and incident response. The paper closes with practical recommendations for model providers, platforms, newsrooms, financial institutions, healthcare systems, legal organizations, regulators, and ordinary users.

Standard Retrieval-Augmented Generation (RAG) chunking methods often create excessive redundancy, increasing storage costs and slowing retrieval. This study explores chunk filtering strategies, such as semantic, topic-based, and named-entity-based methods in order to reduce the indexed corpus while preserving retrieval quality. Experiments are conducted on multiple corpora. Retrieval performance is evaluated using a token-based framework based on precision, recall, and intersection-over-union metrics. Results indicate that entity-based filtering can reduce vector index size by approximately 25% to 36% while maintaining high retrieval quality close to the baseline. These findings suggest that redundancy introduced during chunking can be effectively reduced through lightweight filtering, improving the efficiency of retrieval-oriented components in RAG pipelines.

The evolution of Multimodal Large Language Models (MLLMs) has shifted the focus from text generation to active behavioral execution, particularly via OS agents navigating complex GUIs. However, the transition of these agents into trustworthy daily partners is hindered by a lack of rigorous evaluation regarding safety, efficiency, and multi-modal robustness. Current benchmarks suffer from narrow safety scenarios, noisy trajectory labeling, and limited robustness metrics. To bridge this gap, we propose OS-SPEAR, a comprehensive toolkit for the systematic analysis of OS agents across four dimensions: Safety, Performance, Efficiency, and Robustness. OS-SPEAR introduces four specialized subsets: (1) a S(afety)-subset encompassing diverse environment- and human-induced hazards; (2) a P(erformance)-subset curated via trajectory value estimation and stratified sampling; (3) an E(fficiency)-subset quantifying performance through the dual lenses of temporal latency and token consumption; and (4) a R(obustness)-subset that applies cross-modal disturbances to both visual and textual inputs. Additionally, we provide an automated analysis tool to generate human-readable diagnostic reports. We conduct an extensive evaluation of 22 popular OS agents using OS-SPEAR. Our empirical results reveal critical insights into the current landscape: notably, a prevalent trade-off between efficiency and safety or robustness, the performance superiority of specialized agents over general-purpose models, and varying robustness vulnerabilities across different modalities. By providing a multidimensional ranking and a standardized evaluation framework, OS-SPEAR offers a foundational resource for developing the next generation of reliable and efficient OS agents. The dataset and codes are available at https://github.com/Wuzheng02/OS-SPEAR.

Large language models (LLMs) have achieved strong performance in general machine translation, yet their ability in culture-aware scenarios remains poorly understood. To bridge this gap, we introduce CanMT, a Culture-Aware Novel-Driven Parallel Dataset for Machine Translation, together with a theoretically grounded, multi-dimensional evaluation framework for assessing cultural translation quality. Leveraging CanMT, we systematically evaluate a wide range of LLMs and translation systems under different translation strategy constraints. Our findings reveal substantial performance disparities across models and demonstrate that translation strategies exert a systematic influence on model behavior. Further analysis shows that translation difficulty varies across types of culture-specific items, and that a persistent gap remains between models' recognition of culture-specific knowledge and their ability to correctly operationalize it in translation outputs. In addition, incorporating reference translations is shown to substantially improve evaluation reliability in LLM-as-a-judge, underscoring their essential role in assessing culture-aware translation quality. The corpus and code are available at CanMT.

Large Language Model (LLM)-guided evolutionary search is increasingly used for automated algorithm discovery, yet most current methods track search progress primarily through executable programs and scalar fitness. Even when natural-language reasoning is used through heuristic descriptions or reflection, it typically remains transient mutation context or unstructured memory, rather than organized as persistent population-level state over strategic directions. As a result, evolutionary search can struggle to distinguish syntactically different implementations of the same idea, preserve lower-fitness but strategically promising directions, or detect when an entire family of strategies has saturated. We introduce \model, a modular strategy-space layer that turns language-level strategic reasoning into first-class population-level evolutionary state in LLM-driven program search. \model represents each candidate program with an explicit natural-language strategy, clusters the archive by strategy semantics, retrieves behaviorally complementary inspirations, and periodically navigates the strategy landscape to avoid saturated directions. Without modifying the underlying evolutionary algorithms, \model improves existing evolutionary backbones across algorithm discovery, systems optimization, and agent-scaffold design tasks in most settings. Across four systems benchmarks, \model achieves a 20.6% average relative improvement, with the best single run on Prism scoring 3$\times$ higher. These results suggest that persistent strategy representations provide a practical mechanism for improving the effectiveness and cost-efficiency of LLM-guided evolutionary search, pointing toward compound AI systems whose search capabilities benefit from the structured accumulation and reuse of algorithmic strategies.

Representation learning is fundamental to NLP, but building embeddings that work well at different computational budgets is challenging. Matryoshka Representation Learning (MRL) offers a flexible inference paradigm through nested embeddings; however, learning such structures requires explicit coordination of how information is arranged across embedding dimensionality and model depth. In this work, we propose MIPIC (Matryoshka Representation Learning via Self-Distilled Intra-Relational Alignment and Progressive Information Chaining), a unified training framework designed to produce structurally coherent and semantically compact Matryoshka representations. MIPIC promotes cross-dimensional structural consistency through Self-Distilled Intra-Relational Alignment (SIA), which aligns token-level geometric and attention-driven relations between full and truncated representations using top-k CKA self-distillation. Complementarily, it enables depth-wise semantic consolidation via Progressive Information Chaining (PIC), a scaffolded alignment strategy that incrementally transfers mature task semantics from deeper layers into earlier layers. Extensive experiments on STS, NLI, and classification benchmarks (spanning models from TinyBERT to BGEM3, Qwen3) demonstrate that MIPIC yields Matryoshka representations that are highly competitive across all capacities, with significant performance advantages observed under extreme low-dimensional.

Large Audio-Language Models show consistent performance gains across speech and audio benchmarks, yet high scores may not reflect true auditory perception. If a model can answer questions without processing the acoustic signal, the benchmark fails as a measure of auditory understanding. We present a diagnostic framework using two axes: text prior, which measures answerability from text and general knowledge alone, and audio reliance, which assesses actual dependency on the acoustic signal. Evaluating eight LALMs across three benchmarks, we find that models retain 60-72% of their full audio scores even without any audio input. Moreover, among items that require audio, only 3.0-4.2% need the complete audio clip; the majority can be resolved using localized fragments. These findings challenge the assumption that benchmark performance equals robust audio understanding, and we conclude with practical guidelines for improving evaluation reliability and benchmark design.

Speech-only spoken language models (SLMs) lag behind text and text-speech models in performance, with recent discrete autoregressive (AR) SLMs indicating significant computational and data demands to match text models. Since discretizing continuous speech for AR creates bottlenecks, we explore whether continuous diffusion (CD) SLM is more viable. To quantify the SLMs linguistic quality, we introduce the phoneme Jensen-Shannon divergence (pJSD) metric. Our analysis reveals CD SLMs, mirroring AR behavior, exhibit scaling laws for validation loss and pJSD, and show optimal token-to-parameter ratios decreasing as compute scales. However, for the latter, loss becomes insensitive to choice of data and model sizes, showing potential for fast inference. Scaling CD SLMs to 16B parameters with tens of millions of hours of conversational data enables generation of emotive, prosodic, multi-speaker, multilingual speech, though achieving long-form coherence remains a significant challenge.

As the application of Large Language Models (LLMs) spreads across various industries, there are increasing concerns about the potential for their misuse, especially in sensitive areas such as political discourse. Deliberately aligning LLMs with specific political ideologies, through prompt engineering or fine-tuning techniques, can be advantageous in use cases such as political campaigns, but requires careful consideration due to heightened risks of performance degradation, misinformation, or increased biased behavior. In this work, we propose a multi-dimensional framework inspired by Habermas' Theory of Communicative Action to audit politically aligned language models across four dimensions: effectiveness, fairness, truthfulness, and persuasiveness using automated, quantitative metrics. Applying this to nine popular LLMs aligned via fine-tuning or role-playing revealed consistent trade-offs: while larger models tend to be more effective at role-playing political ideologies and truthful in their responses, they were also less fair, exhibiting higher levels of bias in the form of angry and toxic language towards people of different ideologies. Fine-tuned models exhibited lower bias and more effective alignment than the corresponding role-playing models, but also saw a decline in performance reasoning tasks and an increase in hallucinations. Overall, all of the models tested exhibited some deficiency in at least one of the four metrics, highlighting the need for more balanced and robust alignment strategies. Ultimately, this work aims to ensure politically-aligned LLMs generate legitimate, harmless arguments, offering a framework to evaluate the responsible political alignment of these models.

Long-context ability, has become one of the most important iteration direction of next-generation Large Language Models, particularly in semantic understanding/reasoning, code agentic intelligence and recommendation system. However, the standard softmax attention exhibits quadratic time complexity with respect to sequence length. As the sequence length increases, this incurs substantial overhead in long-context settings, leading the training and inference costs of extremely long sequences deteriorate rapidly. Existing solutions mitigate this issue through two technique routings: i) Reducing the KV cache per layer, such as from the head-level compression GQA, and the embedding dimension-level compression MLA, but the KV cache remains linearly dependent on the sequence length at a 1:1 ratio. ii) Interleaving with KV Cache friendly architecture, such as local attention SWA, linear kernel GDN, but often involve trade-offs among KV Cache and long-context modeling effectiveness. Besides the two technique routings, we argue that there exists an intermediate path not well explored: {Maintaining a linear relationship between the KV cache and sequence length, but performing semantic-level compression through a specific ratio $k$}. This $O(n/k)$ path does not pursue a ``minimum KV cache'', but rather trades acceptable memory costs for complete, referential, and interpretable retention of long distant dependency. Motivated by this, we propose Kwai Summary Attention (KSA), a novel attention mechanism that reduces sequence modeling cost by compressing historical contexts into learnable summary tokens.

Despite the growing use of large language models (LLMs) for writing tasks, users may hesitate to rely on LLMs when personal style is important. Post-editing LLM-generated drafts or translations is a common collaborative writing strategy, but it remains unclear whether users can effectively reshape LLM-generated text to reflect their personal style. We conduct a pre-registered online study ($n=81$) in which participants post-edit LLM-generated drafts for writing tasks where personal style matters to them. Using embedding-based style similarity metrics, we find that post-editing increases stylistic similarity to participants' unassisted writing and reduces similarity to fully LLM-generated output. However, post-edited text still remains stylistically closer in style to LLM text than to participants' unassisted control text, and it exhibits reduced stylistic diversity compared to unassisted human text. We find a gap between perceived stylistic authenticity and model-measured stylistic similarity, with post-edited text often perceived as representative of participants' personal style despite remaining detectable LLM stylistic traces.

Fine-tuning unlocks large language models (LLMs) for specialized applications, but its high computational cost often puts it out of reach for resource-constrained organizations. While cloud platforms could provide the needed resources, data privacy concerns make sharing sensitive information with third parties risky. A promising solution is split learning for LLM fine-tuning, which divides the model between clients and a server, allowing collaborative and secure training through exchanged intermediate data, thus enabling resource-constrained participants to adapt LLMs safely. % In light of this, a growing body of literature has emerged to advance this paradigm, introducing varied model methods, system optimizations, and privacy defense-attack techniques for split learning. To bring clarity and direction to the field, a comprehensive survey is needed to classify, compare, and critique these diverse approaches. This paper fills the gap by presenting the first extensive survey dedicated to split learning for LLM fine-tuning. We propose a unified, fine-grained training pipeline to pinpoint key operational components and conduct a systematic review of state-of-the-art work across three core dimensions: model-level optimization, system-level efficiency, and privacy preservation. Through this structured taxonomy, we establish a foundation for advancing scalable, robust, and secure collaborative LLM adaptation.

Unsupervised automatic readability assessment (ARA) methods have important practical and research applications (e.g., ensuring medical or educational materials are suitable for their target audiences). In this paper, we propose a new zero-shot prompting methodology for ARA and present the first comprehensive evaluation of using large language models (LLMs) as an unsupervised ARA method by testing 10 diverse open-source LLMs (e.g., different sizes and developers) on 14 diverse datasets (e.g., different text lengths and languages). Our findings show that our proposed prompting methodology outperforms prior methods on 13 of the 14 datasets. Furthermore, we propose LAURAE, which combines LLM and readability formula scores to improve robustness by capturing both contextual and shallow (e.g., sentence length) features of readability. Our evaluation demonstrates that LAURAE robustly outperforms prior methods across languages, text lengths, and amounts of technical language.

Multiple myeloma is managed through sequential lines of therapy over years to decades, with each decision depending on cumulative disease history distributed across dozens to hundreds of heterogeneous clinical documents. Whether LLM-based systems can synthesise this evidence at a level approaching expert agreement has not been established. A retrospective evaluation was conducted on longitudinal clinical records of 811 myeloma patients treated at a tertiary centre (2001-2026), covering 44,962 documents and 1,334,677 laboratory values, with external validation on MIMIC-IV. An agentic reasoning system was compared against single-pass retrieval-augmented generation (RAG), iterative RAG, and full-context input on 469 patient-question pairs from 48 templates at three complexity levels. Reference labels came from double annotation by four oncologists with senior haematologist adjudication. Iterative RAG and full-context input converged on a shared ceiling (75.4% vs 75.8%, p = 1.00). The agentic system reached 79.6% concordance (95% CI 76.4-82.8), exceeding both baselines (+3.8 and +4.2 pp; p = 0.006 and 0.007). Gains rose with question complexity, reaching +9.4 pp on criteria-based synthesis (p = 0.032), and with record length, reaching +13.5 pp in the top decile (n = 10). The system error rate (12.2%) was comparable to expert disagreement (13.6%), but severity was inverted: 57.8% of system errors were clinically significant versus 18.8% of expert disagreements. Agentic reasoning was the only approach to exceed the shared ceiling, with gains concentrated on the most complex questions and longest records. The greater clinical consequence of residual system errors indicates that prospective evaluation in routine care is required before these findings translate into patient benefit.

Multi-hop Question Answering (MHQA) aims to answer questions that require multi-step reasoning. It presents two key challenges: generating correct reasoning paths in response to the complex user queries, and accurately retrieving essential knowledge in the face of potential limitations in large language models (LLMs). Existing approaches primarily rely on prompt-based methods to generate reasoning paths, which are further combined with traditional sparse or dense retrieval to produce the final answer. However, the generation of reasoning paths commonly lacks effective control over the generative process, thus leading the reasoning astray. Meanwhile, the retrieval methods over-rely on knowledge matching or similarity scores rather than evaluating the practical utility of the information, resulting in retrieving homogeneous or non-useful information. Therefore, we propose a Structured Entity-Aware Retrieval with Chain-of-Reasoning Navigator framework named SEARCH-R. Specifically, SEARCH-R trains an end-to-end reasoning path navigator, which is able to provide a powerful sub-question decomposer by fine-tuning the Llama3.1-8B model. Moreover, a novel dependency tree-based retrieval is designed to evaluate the informational contribution of the document quantitatively. Extensive experiments on three challenging multi-hop datasets validate the effectiveness of the proposed framework. The code and dataset are available at: https://github.com/Applied-Machine-Learning-Lab/ACL2026_SEARCH-R.

Large language models deployed at runtime can misbehave in ways that clean-data validation cannot anticipate: training-time backdoors lie dormant until triggered, jailbreaks subvert safety alignment, and prompt injections override the deployer's instructions. Existing runtime defenses address these threats one at a time and often assume a clean reference model, trigger knowledge, or editable weights, assumptions that rarely hold for opaque third-party artifacts. We introduce Layerwise Convergence Fingerprinting (LCF), a tuning-free runtime monitor that treats the inter-layer hidden-state trajectory as a health signal: LCF computes a diagonal Mahalanobis distance on every inter-layer difference, aggregates via Ledoit-Wolf shrinkage, and thresholds via leave-one-out calibration on 200 clean examples, with no reference model, trigger knowledge, or retraining. Evaluated on four architectures (Llama-3-8B, Qwen2.5-7B, Gemma-2-9B, Qwen2.5-14B) across backdoors, jailbreaks, and prompt injection (56 backdoor combinations, 3 jailbreak techniques, and BIPIA email + code-QA), LCF reduces mean backdoor attack success rate (ASR) below 1% on Qwen2.5-7B and Gemma-2 and to 1.3% on Qwen2.5-14B, detects 92-100% of DAN jailbreaks (62-100% for GCG and softer role-play), and flags 100% of text-payload injections across all eight (model, domain) cells, at 12-16% backdoor FPR and <0.1% inference overhead. A single aggregation score covers all three threat families without threat-specific tuning, positioning LCF as a general-purpose runtime safety layer for cloud-served and on-device LLMs.

The increasing reliance on Large Language Models (LLMs) across diverse sectors highlights the need for robust domain-specific and language-specific evaluation datasets; however, the collection of such datasets is challenging due to privacy concerns, regulatory restrictions, and the time cost for manual creation. Existing automated benchmarking methods are often limited by relying on pre-existing data, poor scalability, single-domain focus, and lack of multilingual support. We present STELLAR-E - a fully automated system to generate high-quality synthetic datasets of custom size, using minimal human inputs without depending on existing datasets. The system is structured in two stages: (1) We modify the TGRT Self-Instruct framework to create a synthetic data engine that enables controllable, custom synthetic dataset generation, and (2) an evaluation pipeline incorporating statistical and LLM-based metrics to assess the applicability of the synthetic dataset for LLM-based application evaluations. The synthetic datasets reach an average difference of +5.7% in terms of LLM-as-a-judge scores against existing language-specific benchmarks, demonstrating comparable quality for comprehensive assessment of big and small LLMs. While real datasets remain slightly more challenging for LLMs especially for smaller models, this work establishes a scalable and domain-adaptable benchmarking framework that supports fair evaluation of LLM applications, offering a faster alternative to manual approaches and enabling high-efficiency automated quality assurance cycles.

Driving in compliance with traffic laws and regulations is a basic requirement for human drivers, yet autonomous vehicles (AVs) can violate these requirements in diverse real-world scenarios. To encode law compliance into AV systems, conventional approaches use formal logic languages to explicitly specify behavioral constraints, but this process is labor-intensive, hard to scale, and costly to maintain. With recent advances in artificial intelligence, it is promising to leverage large language models (LLMs) to derive legal requirements from traffic laws and regulations. However, without explicitly grounding and reasoning in structured traffic scenarios, LLMs often retrieve irrelevant provisions or miss applicable ones, yielding imprecise requirements. To address this, we propose a novel pipeline that grounds LLM reasoning in a traffic scenario taxonomy through node-wise anchors that encode hierarchical semantics. On Chinese traffic laws and OnSite dataset (5,897 scenarios), our method improves law-scenario matching by 29.1\% and increases the accuracy of derived mandatory and prohibitive requirements by 36.9\% and 38.2\%, respectively. We further demonstrate real-world applicability by constructing a law-compliance layer for AV navigation and developing an onboard, real-time compliance monitor for in-field testing, providing a solid foundation for future AV development, deployment, and regulatory oversight.

Multimodal Retrieval-Augmented Generation (MRAG) addresses key limitations of Multimodal Large Language Models (MLLMs), such as hallucination and outdated knowledge. However, current MRAG systems struggle to distinguish whether retrieved multimodal data truly supports the semantic core of an answer or merely provides superficial relevance. Existing metrics often rely on heuristic position-based confidence, which fails to capture the informational density of multimodal entities. To address this, we propose Multi-modal Evidence Grounding (MEG), a semantic-aware metric that quantifies the contribution of retrieved evidence. Unlike standard confidence measures, MEG utilizes Semantic Certainty Anchoring, focusing on high-IDF information-bearing tokens that better capture the semantic core of the answer. Building on MEG, we introduce MEG-RAG, a framework that trains a multimodal reranker to align retrieved evidence with the semantic anchors of the ground truth. By prioritizing high-value content based on semantic grounding rather than token probability distributions, MEG-RAG improves the accuracy and multimodal consistency of generated outputs. Extensive experiments on the M$^2$RAG benchmark show that MEG-RAG consistently outperforms strong baselines and demonstrates robust generalization across different teacher models.

As large language models (LLMs) evolve into agentic problem solvers, they increasingly rely on external, reusable skills to handle tasks beyond their native parametric capabilities. In existing agent systems, the dominant strategy for incorporating skills is to explicitly enumerate available skills within the context window. However, this strategy fails to scale: as skill corpora expand, context budgets are consumed rapidly, and the agent becomes markedly less accurate in identifying the right skill. To this end, this paper formulates Skill Retrieval Augmentation (SRA), a new paradigm in which agents dynamically retrieve, incorporate, and apply relevant skills from large external skill corpora on demand. To make this problem measurable, we construct a large-scale skill corpus and introduce SRA-Bench, the first benchmark for decomposed evaluation of the full SRA pipeline, covering skill retrieval, skill incorporation, and end-task execution. SRA-Bench contains 5,400 capability-intensive test instances and 636 manually constructed gold skills, which are mixed with web-collected distractor skills to form a large-scale corpus of 26,262 skills. Extensive experiments show that retrieval-based skill augmentation can substantially improve agent performance, validating the promise of the paradigm. At the same time, we uncover a fundamental gap in skill incorporation: current LLM agents tend to load skills at similar rates, regardless of whether a gold skill is retrieved or whether the task actually requires external capabilities. This shows that the bottleneck in skill augmentation lies not only in retrieval but also in the base model's ability to determine which skill to load and when external loading is actually needed. These findings position SRA as a distinct research problem and establish a foundation for the scalable augmentation of capabilities in future agent systems.

Many sign language translation (SLT) systems operate on pose sequences instead of raw video to reduce input dimensionality, improve portability, and partially anonymize signers. The choice of pose estimator is often treated as an implementation detail, with systems defaulting to widely available tools such as MediaPipe Holistic or OpenPose. We present a systematic comparison of pose estimators for pose-based SLT, covering widely used baselines (MediaPipe Holistic, OpenPose) and newer whole-body/high-capacity models (MMPose WholeBody, OpenPifPaf, AlphaPose, SDPose, Sapiens, SMPLest-X). We quantify downstream impact by training a controlled SLT pipeline on RWTH-PHOENIX-Weather 2014 where only the pose representation varies, evaluating with BLEU and BLEURT. To contextualize translation outcomes, we analyze temporal stability, missing hand keypoints, and robustness to occlusion using higher-resolution videos from the Signsuisse dataset. SDPose and Sapiens achieve the best translation performance (BLEU ~11.5), outperforming the common MediaPipe baseline (BLEU ~10). In occlusion cases, Sapiens is correct in all tested instances (15/15), while OpenPifPaf fails in nearly all (1/15) and also yields the weakest translation scores. Estimators that frequently leave out hand keypoints are associated with lower BLEU/BLEURT. We release code that can be used not only to reproduce our experiments, but also considerably lowers the barrier for other researchers to use alternative pose estimators.

Temporal relation classification is the task of determining the temporal relation between pairs of temporal entities in a text. Despite recent advancements in natural language processing, temporal relation classification remains a considerable challenge. Early attempts framed this task using a comprehensive set of temporal relations between events and temporal expressions. However, due to the task complexity, datasets have been progressively simplified, leading recent approaches to focus on the relations between event pairs and to use only a subset of relations. In this work, we revisit the broader goal of classifying interval relations between temporal entities by considering the full set of relations that can hold between two time intervals. The proposed approach, Interval from Point, involves first classifying the point relations between the endpoints of the temporal entities and then decoding these point relations into an interval relation. Evaluation on the TempEval-3 dataset shows that this approach can yield effective results, achieving a temporal awareness score of $70.1$ percent, a new state-of-the-art on this benchmark.

On-device Small Language Models (SLMs) promise fully offline, private AI experiences for mobile users (no cloud dependency, no data leaving the device). But is this promise achievable in practice? This paper presents a longitudinal practitioner case study documenting the engineering challenges of integrating SLMs (Gemma 4 E2B, 2.6B parameters; Qwen3 0.6B, 600M parameters) into Palabrita, a production Android word-guessing game. Over a 5-day development sprint comprising 204 commits (~90 directly AI-related), the system underwent a radical transformation: from an ambitious design where the LLM generated complete structured puzzles (word, category, difficulty, and five hints as JSON) to a pragmatic architecture where curated word lists provide the words and the LLM generates only three short hints, with a deterministic fallback if it fails. We identify five categories of failures specific to on-device SLM integration: output format violations, constraint violations, context quality degradation, latency incompatibility, and model selection instability. For each failure category, we document the observed symptoms, root causes, and the prompt engineering and architectural strategies that effectively mitigated them, including multi-layer defensive parsing, contextual retry with failure feedback, session rotation, progressive prompt hardening, and systematic responsibility reduction. Our findings demonstrate that on-device SLMs are viable for production mobile applications, but only when the developer accepts a fundamental constraint: the most reliable on-device LLM feature is one where the LLM does the least. We distill our experience into eight actionable design heuristics for practitioners integrating SLMs into mobile apps.

The development of practical (multimodal) large language model assistants for Korean weather forecasters is hindered by the absence of a multidimensional, expert-level evaluation framework grounded in authoritative sources. To address this, we introduce K-MetBench, a diagnostic benchmark grounded in national qualification exams. It exposes critical gaps across four dimensions: expert visual reasoning of charts, logical validity via expert-verified rationales, Korean-specific geo-cultural comprehension, and fine-grained domain analysis. Our evaluation of 55 models reveals a profound modality gap in interpreting specialized diagrams and a reasoning gap where models hallucinate logic despite correct predictions. Crucially, Korean models outperform significantly larger global models in local contexts, demonstrating that parameter scaling alone cannot resolve cultural dependencies. K-MetBench serves as a roadmap for developing reliable, culturally aware expert AI agents. The dataset is available at https://huggingface.co/datasets/soyeonbot/K-MetBench .

Long-context reasoning is a critical capability of large language models (LLMs), enabling applications such as long-document understanding, summarization, and code generation. However, efficient autoregressive inference relies on the key-value (KV) cache, whose memory footprint grows linearly with sequence length, leading to a major memory bottleneck. To mitigate this overhead, KV cache pruning methods discard cached tokens with low attention scores during inference. Most existing methods apply a uniform pruning ratio across layers, implicitly assuming that all layers contribute equally to overall model performance. We show that this assumption is suboptimal, as layers differ significantly in their sensitivity to pruning. We propose DepthKV, a layer-dependent pruning framework that allocates a fixed global KV budget across layers based on their sensitivity, rather than using a uniform allocation. Across multiple models and tasks, DepthKV consistently outperforms uniform pruning at the same global pruning ratio, demonstrating more effective utilization of the KV cache budget through layer-dependent allocation.

This paper investigates whether source trustworthiness shapes Turkish evidential morphology and whether large language models (LLMs) track this sensitivity. We study the past-domain contrast between -DI and -mIs in controlled cloze contexts where the information source is overtly external, while only its perceived reliability is manipulated (High-Trust vs. Low-Trust). In a human production experiment, native speakers of Turkish show a robust trust effect: High-Trust contexts yield relatively more -DI, whereas Low-Trust contexts yield relatively more -mIs, with the pattern remaining stable across sensitivity analyses. We then evaluate 10 LLMs in three prompting paradigms (open gap-fill, explicit past-tense gap-fill, and forced-choice A/B selection). LLM behavior is highly model- and prompt-dependent: some models show weak or local trust-consistent shifts, but effects are generally unstable, often reversed, and frequently overshadowed by output-compliance problems and strong base-rate suffix preferences. The results provide new evidence for a trust-/commitment-based account of Turkish evidentiality and reveal a clear human-LLM gap in source-sensitive evidential reasoning.

While Large Language Models (LLMs) have increasingly assisted in historical tasks such as text processing, their capacity for professional-level historical reasoning remains underexplored. Existing benchmarks primarily assess basic knowledge breadth or lexical understanding, failing to capture the higher-order skills, such as evidentiary reasoning,that are central to historical research. To fill this gap, we introduce ProHist-Bench, a novel benchmark anchored in the Chinese Imperial Examination (Keju) system, a comprehensive microcosm of East Asian political, social, and intellectual history spanning over 1,300 years. Developed through deep interdisciplinary collaboration, ProHist-Bench features 400 challenging, expert-curated questions across eight dynasties, accompanied by 10,891 fine-grained evaluation rubrics. Through a rigorous evaluation of 18 LLMs, we reveal a significant proficiency gap: even state-of-the-art LLMs struggle with complex historical research questions. We hope ProHist-Bench will facilitate the development of domain-specific reasoning LLMs, advance computational historical research, and further uncover the untapped potential of LLMs. We release ProHist-Bench at https://github.com/inclusionAI/ABench/tree/main/ProHist-Bench.

Linear activation steering is a powerful approach for eliciting the capabilities of large language models and specializing their behavior using limited labeled data. While effective, existing methods often apply a fixed steering strength to all tokens, resulting in inconsistent steering quality across diverse input prompts. In this work, we introduce Contextual Linear Activation Steering (CLAS), a method that dynamically adapts linear activation steering to context-dependent steering strengths. Across eleven steering benchmarks and four model families, it consistently outperforms standard linear activation steering and matches or exceeds the performance of ReFT and LoRA in settings with limited labeled data. We therefore propose CLAS as a scalable, interpretable, and accurate method for specializing and steering large language models.

Applications based on large language models (LLMs), such as multi-agent simulations, require population diversity among agents. We identify a pervasive failure mode we term \emph{Persona Collapse}: agents each assigned a distinct profile nonetheless converge into a narrow behavioral mode, producing a homogeneous simulated population. To quantify persona collapse, we propose a framework that measures how much of the persona space a population occupies (Coverage), how evenly agents spread across it (Uniformity), and how rich the resulting behavioral patterns are (Complexity). Evaluating ten LLMs on personality simulation (BFI-44), moral reasoning, and self-introduction, we observe persona collapse along two axes: (1) Dimensions: a model can appear diverse on one axis yet structurally degenerate on another, and (2) Domains: the same model may collapse the most in personality yet be the most diverse in moral reasoning. Furthermore, item-level diagnostics reveal that behavioral variation tracks coarse demographic stereotypes rather than the fine-grained individual differences specified in each persona. Counter-intuitively, \textbf{the models achieving the highest per-persona fidelity consistently produce the most stereotyped populations}. We release our toolkit and data to support population-level evaluation of LLMs.

Hybrid sequence models that combine efficient Transformer components with linear sequence modeling blocks are a promising alternative to pure Transformers, but most are still pretrained from scratch and therefore fail to reuse existing Transformer checkpoints. We study upcycling as a practical path to convert pretrained Transformer LLMs into hybrid architectures while preserving short-context quality and improving long-context capability. We call our solution \emph{HyLo} (HYbrid LOng-context): a long-context upcycling recipe that combines architectural adaptation with efficient Transformer blocks, Multi-Head Latent Attention (MLA), and linear blocks (Mamba2 or Gated DeltaNet), together with staged long-context training and teacher-guided distillation for stable optimization. HyLo extends usable context length by up to $32\times$ through efficient post-training and reduces KV-cache memory by more than $90\%$, enabling up to 2M-token prefill and decoding in our \texttt{vLLM} inference stack, while comparable Llama baselines run out of memory beyond 64K context. Across 1B- and 3B-scale settings (Llama- and Qwen-based variants), HyLo delivers consistently strong short- and long-context performance and significantly outperforms state-of-the-art upcycled hybrid baselines on long-context evaluations such as RULER. Notably, at similar scale, HyLo-Qwen-1.7B trained on only 10B tokens significantly outperforms JetNemotron (trained on 400B tokens) on GSM8K, Lm-Harness common sense reasoning and RULER-64K.

Indonesian marketplace reviews mix standard vocabulary with slang, regional loanwords, numeric shorthands, and emoji, making lexicon-based sentiment tools unreliable in practice. This paper describes a two-track classification pipeline applied to the PRDECT-ID dataset, which contains 5,400 product reviews from 29 Indonesian e-commerce categories, each labeled for binary sentiment (Positive/Negative) and five-class emotion (Happy, Sad, Fear, Love, Anger). The first track applies TF-IDF vectorization with a PyCaret AutoML sweep across standard classifiers. The second track is a PyTorch Bidirectional Long Short-Term Memory (BiLSTM) network with a shared encoder and two task-specific output heads. A preprocessing module applies 14 sequential cleaning steps, including a 140-entry slang dictionary assembled from marketplace corpora. Four configurations are benchmarked: BiLSTM Baseline, BiLSTM Improved, BiLSTM Large, and TextCNN. Training uses class-weighted cross-entropy loss, ReduceLROnPlateau scheduling, and early stopping. Both tracks are deployed as Gradio applications on Hugging Face Spaces. Source code is publicly available at https://github.com/ikii-sd/pba2026-crazyrichteam.

Vision-Language-Action (VLA) models are a promising paradigm for generalist robotic manipulation by grounding high-level semantic instructions into executable physical actions. However, prevailing approaches typically adopt a monolithic generation paradigm, directly mapping visual-linguistic features to high-frequency motor commands in a flat, non-hierarchical fashion. This strategy overlooks the inherent hierarchy of robotic manipulation, where complex actions can be naturally modeled in a Hybrid Action Space, decomposing into discrete macro-directional reaching and continuous micro-pose alignment, severely widening the semantic-actuation gap and imposing a heavy representational burden on grounding high-level semantics to continuous actions. To address this, we introduce Libra-VLA, a novel Coarse-to-Fine Dual-System VLA architecture. We explicitly decouple the learning complexity into a coarse-to-fine hierarchy to strike a training equilibrium, while simultaneously leveraging this structural modularity to implement an asynchronous execution strategy. The Semantic Planner predicts discrete action tokens capturing macro-directional intent, while the Action Refiner conditions on coarse intent to generate high-frequency continuous actions for precise alignment. Crucially, our empirical analysis reveals that performance follows an inverted-U curve relative to action decomposition granularity, peaking exactly when the learning difficulty is balanced between the two sub-systems. With the asynchronous design, our approach offers a scalable, robust, and responsive solution for open-world manipulation.

Generating diverse responses is crucial for test-time scaling of large language models (LLMs), yet standard stochastic sampling mostly yields surface-level lexical variation, limiting semantic exploration. In this paper, we propose Exploratory Sampling (ESamp), a decoding approach that explicitly encourages semantic diversity during generation. ESamp is motivated by the well-known observation that neural networks tend to make lower-error predictions on inputs similar to those encountered before, and incur higher prediction error on novel ones. Building on this property, we train a lightweight Distiller at test time to predict deep-layer hidden representations of the LLM from its shallow-layer representations to model the LLM's depth-wise representation transitions. During decoding, the Distiller continuously adapts to the mappings induced by the current generation context. ESamp uses the prediction error as a novelty signal to reweight candidate token extensions conditioned on the current prefix, thereby biasing decoding toward less-explored semantic patterns. ESamp is implemented with an asynchronous training–inference pipeline, with less than 5% worst case overhead (1.2% in the optimized release). Empirical results show that ESamp significantly boosts the Pass@k efficiency of reasoning models, showing superior or comparable performance to strong stochastic and heuristic baselines. Notably, ESamp achieves robust generalization across mathematics, science, and code generation benchmarks and breaks the trade-off between diversity and coherence in creative writing. Our code has released at: https://github.com/LinesHogan/tLLM.

Agent benchmarks remain largely English-centric, while their multilingual versions are often built with machine translation (MT) and limited post-editing. We argue that, for agentic tasks, this minimal workflow can easily break benchmark validity through query-answer misalignment or culturally off-target context. We propose a refined workflow for adapting English benchmarks into multiple languages with explicit functional alignment, cultural alignment, and difficulty calibration using both automated checks and human review. Using this workflow, we introduce GAIA-v2-LILT, a re-audited multilingual extension of GAIA covering five non-English languages. In experiments, our workflow improves agent success rates by up to 32.7% over minimally translated versions, bringing the closest audited setting to within 3.1% of English performance while substantial gaps remain in many other cases. This indicates that a substantial share of the multilingual performance gap is benchmark-induced measurement error, motivating task-level alignment when adapting English benchmarks across languages. The data is available as part of the MAPS package at https://huggingface.co/datasets/Fujitsu-FRE/MAPS/viewer/GAIA-v2-LILT. We also release the code used in our experiments at https://github.com/lilt/gaia-v2-lilt.

Depth pruning improves the inference efficiency of large language models by removing Transformer blocks. Prior work typically treats layer redundancy as an inherent structural property of pretrained networks, emphasizing importance criteria and search algorithms to identify removable layers. In this study, we empirically investigate depth pruning from a functional perspective. Evaluating representative LLM families across diverse calibration configurations and multiple search algorithms, we show that different configurations produce different pruning patterns. Furthermore, under a fixed calibration configuration, complex search algorithms yield marginal performance improvements over simple one-shot methods, converging to similar pruned subsets. Overall, our results suggest that the calibration configuration plays a substantially larger role than the choice of search algorithm in shaping pruning patterns and calibration perplexity, while contributing comparably to variance in downstream reasoning accuracy. This indicates that future pruning efforts may benefit from prioritizing the calibration configuration over search complexity.

Existing web agent benchmarks have largely converged on short, single-site tasks that frontier models are approaching saturation on. However, real world web use consists of long-horizon, multi-site workflows. Common web navigation tasks, such as comparing products across different domains, planning trips across multiple services, or summarizing information from multiple search queries, require sustained context and cross-site reasoning over potentially hours of browsing. To capture and evaluate such behaviors, we introduce Odysseys: a benchmark of 200 long-horizon web tasks derived from real world browsing sessions evaluated on the live Internet. We find that binary pass/fail evaluation is inadequate for long-horizon settings and introduce a rubric-based evaluation, annotating each Odysseys task with an average of 6.1 graded rubrics. We demonstrate that this yields higher agreement with humans and provides a more fine-grained signal than commonly used trajectory-level LLM-as-a-judge evaluation metrics. We tested several leading frontier models and find that the strongest models achieve a success rate of 44.5%, which leaves substantial room for future improvements. Beyond task success, we argue that efficiency is a first-class concern for long-horizon agents. We introduce a Trajectory Efficiency metric (rubric score per step) and find that even frontier agents achieve only 1.15%, marking an evident need for agents that can succeed efficiently and not simply eventually. Odysseys isolates the critical evaluation of long-horizon proficiency in open-web environments, providing a realistic benchmark to measure progress towards computer-use agents that can potentially productively operate for hours. We release our tasks, evaluation scripts, and other results at https://odysseys-website.pages.dev

Clinical abnormality grounding for rare diseases is often hindered by data scarcity, making supervised fine-tuning impractical and single-pass inference highly unstable. We propose Dynamic Decision Learning (DDL), a framework that enables frozen large vision-language models (LVLMs) to refine their decisions across both language and visual spaces by optimizing instructions and consolidating predictions under visual perturbations. This process improves localization quality and produces a consensus-based reliability score that quantifies model confidence. Results on brain imaging benchmarks, including a rare-disease dataset with 281 pathology types across models ranging from 3B to 72B parameters, show that DDL improves mAP@75 by up to 105% on rare-disease cases and outperforms adaptation baselines and supervised fine-tuning. Furthermore, DDL demonstrates stronger calibration between reliability scores and localization accuracy under severe distribution shifts and increasing task difficulty. Code is available at: https://lijunrio.github.io/DDL/

User simulation has long played a vital role in computer science due to its potential to support a wide range of applications. Language, as the primary medium of human communication, forms the foundation of social interaction and behavior. Consequently, simulating conversational behavior has become a key area of study. Recent advancements in large language models (LLMs) have significantly catalyzed progress in this domain by enabling high-fidelity generation of synthetic user conversation. In this paper, we survey recent advancements in LLM-based conversational user simulation. We introduce a novel taxonomy covering user granularity and simulation objectives. Additionally, we systematically analyze core techniques and evaluation methodologies. We aim to keep the research community informed of the latest advancements in conversational user simulation and to further facilitate future research by identifying open challenges and organizing existing work under a unified framework.

Enterprise deep research often fails to produce decision-ready reports due to uneven information coverage, context explosion, and premature stopping. We propose a scalable Enterprise Deep Research (EDR) architecture to address these failures. Our system (i) decomposes requests into coverage-driven objectives via outline generation with reflection, (ii) localizes context with dependency-guided execution and explicit information sharing, and (iii) enforces evidence-based completion criteria so agents iteratively collect information until sufficiency conditions are met. We evaluate on an internal sales enablement task and the public DeepResearch Bench benchmark, where our proposed system design achieves the strongest overall performance compared with competitive deep-research baselines. The results show that dependency-controlled context and explicit evidence sufficiency criteria reduce premature stopping and improve the consistency and depth of enterprise research outputs.

When an LLM formalizes natural language, how do we know the output is faithful? We propose a roundtrip verification approach which does not require ground-truth annotations: formalize a statement, translate the result back to natural language, re-formalize, and use a formal tool to check logical equivalence. When the two formalizations agree, this provides evidence of a faithful formalization. When they disagree, a stage-level diagnosis localizes the error to a specific translation step, and a scoped repair operator attempts to correct that step. We evaluate the framework on two statutory domains (the Texas Transportation Code and the Texas Parks and Wildlife Code) using two LLMs (Claude Opus~4.6 and GPT-5.2) with three repair baselines. Diagnosis-guided scoped repair is the most effective method, with effectiveness contingent on the reliability of the diagnosis function. Across both domains and both models, under our full repair system, rules that fail the equivalence check show 1.4x-2.5x more NLI drift than rules that pass it.

While current Large Language Models (LLMs) exhibit remarkable reasoning capabilities through test-time compute scaling (TTS), their massive parameter counts and high inference costs have motivated the development of pruning methods that can reduce model size without sacrificing performance. However, specific to reasoning LLMs, prior work has shown that structured pruning (methods which removes entire set of layer blocks), significantly degrades TTS reasoning performance. In this work, we revisit this assumption and instead investigate whether unstructured pruning (methods that carefully remove only certain redundant/detrimental weights) exhibits similar limitations. Surprisingly, our extensive experiments across four reasoning benchmarks on two reasoning LLMs: s1.1-7B and Qwen3-8B, consistently show that unstructured pruning augments TTS performance compared to structured pruning, and at times can even outperform the unpruned full-weight LLMs. Furthermore, we also empirically study the impact of different layer-wise sparsity allocation strategies, which are an important parametric choice for instantiating unstructured pruning methods. These findings challenge the conventional notion that pruning always reduces TTS performance and in fact, suggest that carefully undertaken pruning can improve TTS effectiveness even further.

Korean aegyo is a socially recognized childlike speaking style used predominantly in romantic interactions among adults. This study examined vowel space modification in aegyo by analyzing formant frequencies from twelve Seoul Korean speakers who produced identical scripts in aegyo and non-aegyo styles. Results show that aegyo speech features a significant increase in F1 values across vowels and selective fronting of front vowels, leading to vowel space expansion but mainly a shift to higher F1. These findings suggest that adult speakers stylize childlike speech by imitating the shorter vocal tract of children, mainly through global vowel lowering and partial fronting.

Large Language Models are being increasingly deployed as the decision-making core of autonomous agents capable of effecting change in external environments. Yet, in conversational benchmarks, which simulate real-world customer-centric issue resolution scenarios, these agents frequently fail due to the cascading effects of incorrect decision-making. These challenges are particularly pronounced for open-source LLMs with smaller parameter sizes, limited context windows, and constrained inference budgets, which contribute to increased error accumulation in agentic settings. To tackle these challenges, we present the Failure-Aware Meta-Agentic (FAMA) framework. FAMA operates in two stages: first, it analyzes failure trajectories from baseline agents to identify the most prevalent errors; second, it employs an orchestration mechanism that activates a minimal subset of specialized agents tailored to address these failures by injecting a targeted context for the tool-use agent before the decision-making step. Experiments across open-source LLMs demonstrate performance gains up to 27% across evaluation modes over standard baselines. These results highlight that targeted curation of context through specialized agents to address common failures is a valuable design principle for building reliable, multi-turn tool-use LLM agents that simulate real-world conversational scenarios.

A multilingual collection may contain useful knowledge in other languages to supplement and correct the facts in the original language for Retrieval-Augmented Generation (RAG). However, the vanilla approach that simply concatenates multiple pieces of knowledge from different languages into the context may fail to improve effectiveness due to the potential disparities across languages. To better leverage multilingual knowledge, we propose CroSearch-R1, a search-augmented reinforcement learning framework to integrate multilingual knowledge into the Group Relative Policy Optimization (GRPO) process. In particular, the approach adopts a multi-turn retrieval strategy with cross-lingual knowledge integration to dynamically align the knowledge from other languages as supplementary evidence into a unified representation space. Furthermore, we introduce a multilingual rollout mechanism to optimize reasoning transferability across languages. Experimental results demonstrate that our framework effectively leverages cross-lingual complementarity and improves the effectiveness of RAG with multilingual collections.

Vision-language models (VLMs) are increasingly used as automated judges for multimodal systems, yet their scores provide no indication of reliability. We study this problem through conformal prediction, a distribution-free framework that converts a judge's point score into a calibrated prediction interval using only score-token log-probabilities, with no retraining. We present the first systematic analysis of conformal prediction for VLM-as-a-Judge across 3 judges and 14 visual task categories. Our results show that evaluation uncertainty is strongly task-dependent: intervals cover ~40% of the score range for aesthetics and natural images but expand to ~70% for chart and mathematical reasoning, yielding a quantitative reliability map for multimodal evaluation. We further identify a failure mode not captured by standard evaluation metrics, ranking-scoring decoupling, where judges achieve high ranking correlation while producing wide, uninformative intervals, correctly ordering responses but failing to assign reliable absolute scores. Finally, we show that interval width is driven primarily by task difficulty and annotation quality, i.e., the same judge and method yield 4.5x narrower intervals on a clean, multi-annotator captioning benchmark. Code: https://github.com/divake/VLM-Judge-Uncertainty

Multimodal Large Language Models (MLLMs) have shown transformative potential in medical applications, yet their performance is hindered by conventional data curation strategies that rely on coarse-grained partitioning by modality or department. Such fragmented approaches fail to capture the hierarchical and interconnected nature of clinical medical knowledge, limiting the models' ability to perform fine-grained recognition and complex reasoning. In this paper, we propose a novel Entity-Centric Medical Data Engineering framework. We automatically extract entities from authoritative medical literature to construct a Medical Entity Tree (MET), a hierarchical structure that systematically encodes diseases, anatomical structures, modalities, and symptoms into a unified knowledge repository. Building upon the MET, we propose an advanced data engine that includes: (1) node-guided retrieval to anchor raw data to specific medical concepts, (2) a two-stage hybrid filtering and alignment pipeline to ensure precise visual-semantic correspondence, and (3) knowledge-aware data synthesis to generate enriched captions and targeted reasoning VQA pairs, leveraging structural constraints. Extensive evaluations across six medical benchmarks demonstrate that our approach significantly enhances the medical capabilities of general-purpose MLLMs, improving their ability to handle complex clinical queries and achieve state-of-the-art performance in diverse medical contexts.

In recent years, the rapid proliferation of open-source large language models (LLMs) has spurred efforts to turn general-purpose models into domain specialists. However, many domain-specialized LLMs are developed using datasets and training protocols that are not aligned with the nuanced requirements of real-world applications. In the legal domain, where precision and reliability are essential, this lack of consideration limits practical utility. In this study, we propose a systematic training framework grounded in the practical needs of the legal domain, with a focus on Korean law. We introduce LegalMidm, a Korean legal-domain LLM, and present a methodology for constructing high-quality, use-case-driven legal datasets and optimized training pipelines. Our approach emphasizes collaboration with legal professionals and rigorous data curation to ensure relevance and factual accuracy, and demonstrates effectiveness in key legal tasks.

Retrieval-Augmented Generation (RAG) models frequently produce answers grounded in parametric memory rather than the retrieved context, undermining the core promise of retrieval augmentation. A fundamental obstacle to fixing this unfaithfulness is the lack of training data that explicitly requires models to prefer context over internal knowledge. We introduce Faithfulness-QA, a large-scale dataset of 99,094 samples constructed through counterfactual entity substitution. Starting from two established extractive QA benchmarks–SQuAD and TriviaQA–we automatically identify answer-bearing named entities in each context, replace them with type-consistent alternatives drawn from a curated bank of 76,953 entities, and thereby manufacture controlled knowledge conflicts between context and parametric memory. Rigorous quality filtering ensures 100% pass rates across four automated checks on random 200-sample audits. We release the full dataset, the construction pipeline, and a typed entity bank covering eight named entity categories. Faithfulness-QA is designed as a training resource for attention-based faithfulness objectives and as an evaluation benchmark for measuring context-grounding behavior in RAG systems. Data and code are available at https://github.com/qzhangFDU/faithfulness-qa-dataset.

Cutscenes are carefully choreographed cinematic sequences embedded in video games and interactive media, serving as the primary vehicle for narrative delivery, character development, and emotional engagement. Producing cutscenes is inherently complex: it demands seamless coordination across screenwriting, cinematography, character animation, voice acting, and technical direction, often requiring days to weeks of collaborative effort from multidisciplinary teams to produce minutes of polished content. In this work, we present Cutscene Agent, an LLM agent framework for automated end-to-end cutscene generation. The framework makes three contributions: (1)~a Cutscene Toolkit built on the Model Context Protocol (MCP) that establishes \emph{bidirectional} integration between LLM agents and the game engine – agents not only invoke engine operations but continuously observe real-time scene state, enabling closed-loop generation of editable engine-native cinematic assets; (2)~a multi-agent system where a director agent orchestrates specialist subagents for animation, cinematography, and sound design, augmented by a visual reasoning feedback loop for perception-driven refinement; and (3)~CutsceneBench, a hierarchical evaluation benchmark for cutscene generation. Unlike typical tool-use benchmarks that evaluate short, isolated function calls, cutscene generation requires long-horizon, multi-step orchestration of dozens of interdependent tool invocations with strict ordering constraints – a capability dimension that existing benchmarks do not cover. We evaluate a range of LLMs on CutsceneBench and analyze their performance across this challenging task.

Large Language Models are increasingly being deployed to extract structured data from unstructured and semi-structured sources: parsing invoices, medical records, and converting PDF documents to database entries. Yet existing benchmarks for structured output generation either focus on schema compliance alone, or evaluate value correctness within a single source domain. We introduce SOB (The Structured Output Benchmark), a multi-source benchmark spanning three source modalities: native text, images, and audio conversations. All models receive a text-normalized representation of their context regardless of source modality; this deliberate design isolates structured-output capability from raw vision or speech-processing quality, ensuring a fair, source-agnostic comparison. Our benchmark comprises 5,000 text evaluation records derived from multi-hop QA drawn from a 25,091-record full corpus, 209 image records from OCR-processed PDFs across seven document types including multi-column layouts, dense tables, scanned historical documents, small-print text, and mathematical typesetting, and 115 audio records from the AMI corpus. Each record pairs a natural-language question with a JSON schema that the model must follow and a ground-truth answer verified against the source context. We evaluate 21 frontier and open-weight models across three source domains and seven metrics. Our results reveal a consistent pattern: models achieve near-perfect schema compliance, yet the best Value Accuracy, measured by exact leaf-value match, reaches only 83.0% on text, 67.2% on images, and 23.7% on audio, where longer context makes extraction substantially harder. We release the dataset, evaluation pipeline, and all related code.

\textbf{Background:} Dutch medical corpora are scarce, limiting NLP development. \\ \textbf{Methods:} We translated English datasets, identified medical text in generic corpora, and extracted open Dutch medical resources. \\ \textbf{Results:} The resulting corpus comprises $\pm$ 35 billion tokens across the medical domain in about 100 million documents, freely available on Hugging Face. \\ \textbf{Conclusion:} This work establishes the first large-scale Dutch medical language corpus for pre-training and downstream NLP tasks.

This paper presents a pipeline designed to transform raw Wikimedia dumps into quality textual corpora for seven South Slavic languages. The work is divided into two major phases. The first involves extracting and cleaning text from raw dumps of Wikipedia, Wikisource, Wikibooks, Wikinews, and Wikiquote. This step requires careful handling of raw wiki markup to isolate, first of all, textual articles, and then usable natural language text within them. The second phase addresses the challenge of questionable or low-quality articles, which are often generated from databases or structured knowledge bases. These articles are characterised by repetitive patterns, generic phrasing, and minimal to no original content. To mitigate their impact, a n-gram-based filtering strategy was employed to detect high levels of textual redundancy between articles and then remove such articles from the corpora entirely. The resulting datasets aim to provide linguistically rich texts suitable for training language models or conducting comparative research across South Slavic languages. By combining systematic extraction with quality control, this work contributes to the creation of reliable, high-information corpora that reflect the authentic cultural contexts of languages. While focused on the South Slavic case in the paper, the approach is mostly language-agnostic and can be generalised to other languages.

This paper benchmarks a classical machine learning approach based on PyCaret AutoML against a deep learning approach based on IndoBERT fine-tuning for binary sentiment analysis of Indonesian-language Twitter comments related to Ibu Kota Nusantara (IKN). The dataset contains 1,472 manually labeled samples, consisting of 780 negative and 692 positive comments. In the machine learning setting, Logistic Regression, Naive Bayes, and Support Vector Machine were evaluated using 10-fold cross-validation, with Logistic Regression achieving the best performance among the classical models at 77.57% accuracy and 77.17% F1-score. In the deep learning setting, the indobenchmark/indobert-base-p1 model was fine-tuned for five epochs and achieved 89.59% test accuracy and 89.37% F1-score. The results show that IndoBERT substantially outperforms the machine learning baselines, highlighting the effectiveness of Transformer-based contextual representations for informal Indonesian social media text.

Do large language models (LLMs) truly acquire embodied cognition and cultural conventions from text? We introduce demonstratives, fundamental spatial expressions like "this/that" in English and "zhè/nà" in Chinese, as a novel probe for grounded knowledge. Using 6,400 responses from 320 native speakers, we establish a human baseline: English speakers reliably distinguish proximal-distal referents but struggle with perspective-taking, while Chinese speakers switch perspectives fluently but tolerate distal ambiguity. In contrast, five state-of-the-art LLMs fail to inherently understand the proximal-distal contrast and show no cultural differences, defaulting to English-centric reasoning. Our study contributes (i) a new task, based on demonstratives, as a new lens for evaluating embodied cognition and cultural conventions; (ii) empirical evidence of cross-cultural asymmetries in human interpretation; (iii) a new perspective on the egocentric-sociocentric debate, showing both orientations coexist but vary across languages; and (iv) a call to address individual variation in future model design.

Commercial TTS systems produce near-native Indic audio, but the best open-source bases (Chatterbox, Indic Parler-TTS, IndicF5) trail them on measured phonological dimensions, and the most widely adopted multilingual base (Chatterbox, 23 languages) does not even tokenise Telugu or Tamil. We ask: what is the minimum intervention that brings such a non-Indic-native base to commercial-class output on Telugu, Tamil, and Hindi, without training a new acoustic decoder and without any commercial TTS training data? We combine three pieces: (1) BUPS, a Brahmic Unified Phoneme Space that deterministically romanises seven Indic scripts to ISO-15919 so Chatterbox's Latin tokeniser can process them; (2) a LoRA adapter on only the text-token predictor (Chatterbox's t3), trained on ~1,220h of licensed Indic audio with a Hindi-proxy language_id; (3) a voice-prompt recovery recipe – an 8-11s same-language reference clip plus three sampling overrides (exaggeration 0.7, temperature 0.6, min_p 0.1; "Config B") – that recovers commercial-class acoustic output with no acoustic-decoder training. On Hindi, the LoRA regresses accuracy and we instead use vanilla Chatterbox + Config B, giving a two-branch deployment. Evaluated on 10-utterance pilot sets with the companion PSP benchmark, Praxy Voice matches or slightly leads commercial baselines: 26.7% retroflex collapse on Telugu (vs Sarvam Bulbul 33.3%), 71% Tamil-zha collapse (vs commercial trio's 86%), 0.025 LLM-WER on Hindi (tied with Cartesia Sonic-3). For intra-sentential code-mix we add a third branch (IndicF5 + native-script transliteration) that drops code-mix LLM-WER from 0.80-0.85 to 0.14-0.27 across Hi/Te/Ta. We release R6 LoRA weights (Apache-2.0), inference code and router (MIT), and a Gradio demo.

Navigating AI regulation across jurisdictions is increasingly difficult for policymakers, legal professionals, and researchers. To address this, we present a multi-jurisdictional Retrieval-Augmented Generation system for global AI regulation. Our corpus includes 242 documents across 68 jurisdictions, ranging from formal legislation like the EU AI Act to unstructured policy documents such as national AI strategies. The system makes three technical contributions: type-specific chunking that preserve legal structure across heterogenous documents; conditional retrieval routing with entity detection and metadata for legal citations; and priority-based re-ranking to boost enacted legislation over policy and secondary sources. Evaluation of 50 queries reveals strong performance across both single-entity and multi-jurisdictional questions, achieving 0.87 average faithfulness and 0.84 average answer relevancy. Single-entity queries achieve 0.86 average faithfulness and 0.92 average answer relevancy, while multi-jurisdictional comparison queries achieve 0.88 average faithfulness and 0.75 average answer relevancy. These findings highlight the effectiveness of domain-specific retrieval strategies for navigating complex, heterogenous regulatory corpora.

Sentiment analysis of product reviews on e-commerce platforms plays a critical role in automatically understanding customer satisfaction and providing actionable insights for sellers seeking to improve product quality. This paper presents a comprehensive benchmarking study comparing a Machine Learning (ML) approach via the PyCaret AutoML framework against a Deep Learning (DL) approach based on a Bidirectional Long Short-Term Memory (BiLSTM) architecture with an Attention mechanism for binary sentiment classification on Indonesian product reviews. The dataset comprises 19,728 samples balanced equally between positive and negative reviews. For the ML approach, three prominent algorithms were evaluated via 10-fold stratified cross-validation: Logistic Regression (LR), Support Vector Machine (SVM) with a linear kernel, and Light Gradient Boosting Machine (LightGBM). Logistic Regression achieved the best ML performance with an accuracy of 97.26\% and an F1-score of 97.26\%. The BiLSTM with Attention model, evaluated on 3,946 held-out test samples, achieved an accuracy of 97.24\% and an F1-score of 97.24\%. These comparative results demonstrate that traditional ML algorithms with proper preprocessing and feature extraction can compete closely with, and even marginally outperform, more complex sequential DL architectures on high-dimensional datasets, while simultaneously offering greater computational efficiency.

We investigate linguistic biases in LLM-based restaurant and product recommendations given prompts varying across Southern American English (AE), Indian English (IE), and Code-Switched Hindi-English dialects, using the Yelp Open dataset (Yelp Inc., 2023) and Walmart product reviews dataset (PromptCloud,2020). We add lists of restaurant and product names balanced by cuisine type and product category to the prompts given to the LLM, and we zero-shot prompt the LLMs in a cold-start setting to select the top-20 restaurant and product recommendations from these lists for each of the dialect-varied prompts. We prompt LLMs using different list samples across 20 seeds for better generalization, and aggregate per cuisine-type and per category response counts for each seed, question/prompt, and LLM model. We run mixed-effects regression models for each model family and topic (restaurant/product) with the aggregate response counts as the dependent, and conduct likelihood ratio tests for the fixed effects with post-hoc pairwise testing of estimated marginal means differences, to investigate group-level differences in recommendation counts by model size and dialect type. Results show that dialect plays a role in the type of restaurant selected across the models tested with the mistral-small-3.1 model and both the llama-3.1 family models tested showing more sensitivity to Indian English and Code-Switched prompts. In terms of product recommendations, the llama-3.1-70B-model is particularly sensitive to Code-Switched prompts in four out of seven categories, and more beauty and home category recommendations are seen when using the Indian English and Code-Switched prompts for larger and smaller models, respectively. No broad trends are seen in the model-size based differences, with differing recommendations based on model sizes conditioned by the type of dialect.

Standard text-to-speech (TTS) evaluation measures intelligibility (WER, CER) and overall naturalness (MOS, UTMOS) but does not quantify accent. A synthesiser may score well on all four yet sound non-native on features that are phonemic in the target language. For Indic languages, these features include retroflex articulation, aspiration, vowel length, and the Tamil retroflex approximant (letter zha). We present PSP, the Phoneme Substitution Profile, an interpretable, per-phonological-dimension accent benchmark for Indic TTS. PSP decomposes accent into six complementary dimensions: retroflex collapse rate (RR), aspiration fidelity (AF), vowel-length fidelity (LF), Tamil-zha fidelity (ZF), Frechet Audio Distance (FAD), and prosodic signature divergence (PSD). The first four are measured via forced alignment plus native-speaker-centroid acoustic probes over Wav2Vec2-XLS-R layer-9 embeddings; the latter two are corpus-level distributional distances. In this v1 we benchmark four commercial and open-source systems (ElevenLabs v3, Cartesia Sonic-3, Sarvam Bulbul, Indic Parler-TTS) on Hindi, Telugu, and Tamil pilot sets, with a fifth system (Praxy Voice) included on all three languages, plus an R5->R6 case study on Telugu. Three findings: (i) retroflex collapse grows monotonically with phonological difficulty Hindi < Telugu < Tamil (~1%, ~40%, ~68%); (ii) PSP ordering diverges from WER ordering – commercial WER-leaders do not uniformly lead on retroflex or prosodic fidelity; (iii) no single system is Pareto-optimal across all six dimensions. We release native reference centroids (500 clips per language), 1000-clip embeddings for FAD, 500-clip prosodic feature matrices for PSD, 300-utterance golden sets per language, scoring code under MIT, and centroids under CC-BY. Formal MOS-correlation is deferred to v2; v1 reports five internal-consistency signals plus a native-audio sanity check.

Large Language Models (LLMs) have shown strong potential for narrative generation, but their use in complex, multi-layered role-playing game (RPG) worlds is still limited by issues of coherence, controllability, and structural consistency. This paper explores a dependency-aware, multi-stage prompt pipeline for procedural RPG content generation that models narrative dependencies through structured intermediate representations. The approach decomposes generation into sequential stages: world building, non-player character creation, player character creation, campaign-level quest planning, and quest expansion. Each stage conditions on structured JSON outputs from previous stages. By enforcing schemas and explicit data flow, the pipeline reduces narrative drift, limits hallucinations, and supports scalable creation of interconnected narrative elements. The system is evaluated qualitatively through human-centered analysis across multiple independent runs. Outputs are assessed using criteria such as structural completeness, internal consistency, narrative coherence, diversity, and actionability. Results show that the pipeline consistently generates logically sound and structurally valid RPG content, without quality degradation as complexity increases. Separating high-level campaign planning from detailed quest expansion improves both global structure and local storytelling. These findings suggest that dependency-aware prompt pipelines with structured intermediate representations are an effective design pattern for LLM-based procedural content generation. This approach may also generalize to other domains requiring sequential reasoning over evolving contextual states.

Large Language Models (LLMs) are increasingly used not only for instrumental tasks, but as always-available and non-judgmental confidants for emotional support. Yet what drives adoption and how users perceive emotional support interactions across countries remains unknown. To address this gap, we present the first large-scale cross-cultural study of LLM use for emotional support, surveying 4,641 participants across seven countries (USA, UK, Germany, France, Spain, Italy, and The Netherlands). Our results show that adoption rates vary dramatically across countries (from 20% to 59%). Using mixed models that separate cultural effects from demographic composition, we find that: Being aged 25-44, religious, married, and of higher socioeconomic status are predictors of positive perceptions (trust, usage, perceived benefits), with socioeconomic status being the strongest. English-speaking countries consistently show more positive perceptions than Continental European countries. We further collect a corpus of 731 real multilingual prompts from user interactions, showing that users mainly seek help for loneliness, stress, relationship conflicts, and mental health struggles. Our findings reveal that LLM emotional support use is shaped by a complex sociotechnical landscape and call for a broader research agenda examining how these systems can be developed, deployed, and governed to ensure safe and informed access.

We present Marco-MoE, a suite of fully open multilingual sparse Mixture-of-Experts (MoE) models. Marco-MoE features a highly sparse design in which only around 5\% of the total parameters are activated per input token. This extreme sparsity, combined with upcycling from dense models, enables efficient pre-training on 5T tokens. Our models surpass similarly-sized competitors on English and multilingual benchmarks, achieving a best-in-class performance-to-compute ratio. We further post-train these models to create Marco-MoE-\textsc{Instruct} variants, which surpass the performance of competing models possessing $3$–$14\times$ more activated parameters. Our analysis reveals that Marco-MoE learns structured expert activation patterns shared across related languages, while maintaining highly specialized utilization for linguistically isolated ones. We further show that Marco-MoE allows for scalable language expansion without the interference typical of dense models. To support the community, we disclose our full training datasets, recipes, and model weights.

The closure of Perspective API at the end of 2026 discards what has functioned as the de facto standard for automated toxicity measurement in NLP, CSS, and LLM evaluation research. We document the structural dependence that the communities built on this single proprietary tool and discuss how this dependence caused epistemic problems that have affected - and will likely continue to affect - collective research efforts. Perspective's model was periodically updated without versioning or disclosure, its annotation structure reflected a single corporate operationalisation of a contested concept, and its scores were used simultaneously as an evaluation target and an evaluation standard. Its closure leaves behind non-updatable benchmarks, irreproducible results, and ultimately a field at risk of perpetuating these issues by turning to closed-source LLMs. We use Perspective's announced termination as an opportunity to call for an independent, valid, adaptable, and reproducible toxicity and hate speech measurement infrastructure, with the technical and governance requirements outlined in this paper.

Recent audio-aware large language models (ALLMs) have demonstrated strong capabilities across diverse audio understanding and reasoning tasks, but they still frequently produce hallucinated or overly confident outputs. While uncertainty estimation has been extensively studied in text-only LLMs, it remains largely unexplored for ALLMs, where audio-conditioned generation introduces additional challenges such as perceptual ambiguity and cross-modal grounding. In this work, we present the first systematic empirical study of uncertainty estimation in ALLMs. We benchmark five representative methods, including predictive entropy, length-normalized entropy, semantic entropy, discrete semantic entropy, and P(True), across multiple models and diverse evaluation settings spanning general audio understanding, reasoning, hallucination detection, and unanswerable question answering. Our results reveal two key findings. First, semantic-level and verification-based methods consistently outperform token-level baselines on general audio reasoning benchmarks. Second, on trustworthiness-oriented benchmarks, the relative effectiveness of uncertainty methods becomes notably more model- and benchmark-dependent, indicating that conclusions drawn from general reasoning settings do not straightforwardly transfer to hallucination and unanswerable-question scenarios. We further explore uncertainty-based adaptive inference as a potential downstream application. We hope this study provides a foundation for future research on reliable, uncertainty-aware audio-language systems.

Real-time automatic speech recognition (ASR) systems face a fundamental trade-off between transcription accuracy and computational efficiency, particularly when deploying large-scale transformer models like Whisper. Existing streaming approaches either sacrifice accuracy through aggressive chunking or incur prohibitive memory costs through unbounded context accumulation. We present WhisperPipe, a novel streaming architecture that achieves bounded memory consumption while maintaining transcription quality through three key innovations a hybrid Voice Activity Detection (VAD) pipeline combining Silero VAD with energy-based filtering to reduce false activations by 34%, a dynamic buffering mechanism with overlapping context windows that prevents information loss at segment boundaries, and an adaptive processing strategy that balances latency and accuracy based on speech characteristics. Evaluated on 2.5 hours of diverse audio data, WhisperPipe demonstrates a median end-to-end latency of 89ms (90th percentile: 142ms) while consuming 48% less peak GPU memory and 80.9% lower average GPU utilization compared to baseline Whisper implementations. The system maintains stable memory usage over extended sessions, with zero growth rate across 150-minute continuous operation. Comparative analysis against related work shows that WhisperPipe achieves competitive accuracy (WER within 2% of offline Whisper) while operating at 3-5x lower latency than existing streaming solutions. The architecture's modular design enables deployment across resource-constrained environments, from edge devices to cloud infrastructure. Our results demonstrate that careful architectural design can reconcile the competing demands of real-time responsiveness and model sophistication in production ASR systems.

We report a striking statistical regularity in frontier LLM outputs that enables a CPU-only scoring primitive running at 2.6 microseconds per token, with estimated latency up to 100,000$\times$ (five orders of magnitude) below existing sampling-based detectors. Across six contemporary models from five independent vendors, two generation sizes, and five held-out domains, token rank-frequency distributions converge to the same two-parameter Mandelbrot ranking distribution, with 34 of 36 model-by-domain fits exceeding $R^{2} = 0.94$ and 35 of 36 favoring Mandelbrot over Zipf by AIC. The shared family does not collapse the models into statistical duplicates. Fitted Mandelbrot parameters remain cleanly separable between models: the cross-model spread in $q$ (1.63 to 3.69) exceeds its per-model bootstrap standard deviation (0.03 to 0.10) by more than an order of magnitude, yielding tens of standard deviations of separation per few thousand output tokens. Two capabilities follow. First, statistical model fingerprinting: text from a vendor-delivered LLM can be tested against its claimed model family without cryptographic watermarks or access to model internals, supporting provenance verification and silent-substitution audits. Second, a model-agnostic reference distribution for black-box output assessment, from which we derive a single-pass scoring primitive that composes with model log probabilities when available and degrades to a rank-only mode usable on closed APIs. Pilot results on FRANK, TruthfulQA, and HaluEval map where the primitive helps (lexical anomalies, unsupported entities) and where it structurally cannot (reasoning errors in domain-appropriate vocabulary). We position the primitive as a first-pass triage layer in compound evaluation stacks, not as a replacement for sampling-based or source-conditioned verifiers.

Although the cultural (mis)alignment of Large Language Models (LLMs) has attracted increasing attention – often framed in terms of cultural bias – until recently there has been limited work on the design and development of datasets for cultural assessment. Here, we review existing approaches to such datasets and identify their main limitations. To address these issues, we propose design guidelines for annotators and report on the construction of a dataset built according to these principles. We further present a series of contrastive experiments conducted with this dataset. The results demonstrate that our design yields test sets with greater discriminative power, effectively distinguishing between models specialized for a given culture and those that are not, ceteris paribus.

Reliable evaluation of large language model (LLM)-generated summaries remains an open challenge, particularly across heterogeneous domains and document lengths. We conduct a comprehensive meta-evaluation of 14 automatic summarization metrics and LLM-based evaluators across seven datasets spanning five domains, covering documents from short news articles to long scientific, governmental, and legal texts (2K-27K words) with over 1,500 human-annotated summaries. Our results show that traditional lexical overlap metrics (e.g., ROUGE, BLEU) exhibit weak or negative correlation with human judgments, while task-specific neural metrics and LLM-based evaluators achieve substantially higher alignment, especially for linguistic quality assessment. Leveraging these findings, we propose LLM-ReSum, a self-reflective summarization framework that integrates LLM-based evaluation and generation in a closed feedback loop without model finetuning. Across three domains, LLM-ReSum improves low-quality summaries by up to 33% in factual accuracy and 39% in coverage, with human evaluators preferring refined summaries in 89% of cases. We additionally introduce PatentSumEval, a new human-annotated benchmark for legal document summarization comprising 180 expert-evaluated summaries. All code and datasets will be released in GitHub.

Modeling the emergence of human-like lexicons in computational systems has advanced through the use of interacting neural agents, which simulate both learning and communicative pressures. The NeLLCom-Lex framework (Zhang et al., 2025) allows neural agents to develop pragmatic color naming behavior and human-like lexicons through supervised learning (SL) from human data and reinforcement learning (RL) in referential games. Despite these successes, the lexicons that emerge diverge systematically from human color categories, producing highly non-convex regions in color space, which contrast with the convexity typical of human categories. To address this, we introduce two factors, upsampling rare color terms during SL and multi-listener RL interactions, and adopt a convexity measure to quantify geometric coherence. We find that upsampling improves lexical diversity and system-level informativeness of the color lexicon, while many-listener setups promote more convex color categories. The combination of moderate upsampling and multiple listeners produces lexicons most similar to human systems.

Multilingual retrieval-augmented generation (mRAG) is often implemented within a fixed retrieval space, typically via query or document translation or multilingual embedding vector representations. However, this approach may be inadequate for culturally grounded queries, in which retrieval-condition misalignment may occur. Even strong retrievers and generators may struggle to produce culturally relevant answers when sourcing evidence from inappropriate linguistic or regional contexts. To this end, we introduce CORAL (COntext-aware Retrieval with Agentic Loop, an adaptive retrieval methodology for mRAG that enables iterative refinement of both the retrieval space (corpora) and the retrieval probe (query) based on the quality of the evidence. The overall process includes: (1) selecting corpora, (2) retrieving documents, (3) critiquing evidence for relevance and cultural alignment, and (4) checking sufficiency. If the retrieved documents are insufficient to answer the query correctly, the system (5) reselects corpora and rewrites the query. Across two cultural QA benchmarks, CORAL achieves up to a 3.58%p accuracy improvement on low-resource languages relative to the strongest baselines.

Contemporary neural machine translation (NMT) systems are almost exclusively built by training on supervised parallel data. Despite the tremendous progress achieved, these systems still exhibit persistent translation errors. This paper proposes that a post-training paradigm based on reinforcement learning (RL) can effectively rectify such mistakes. We introduce a novel framework that requires only a general text corpus and an expert translator which can be either human or an AI system to provide iterative feedback. In our experiments, we focus specifically on English-to-German translation as a representative high-resource language pair. Crucially, we implement this RL-based post-training using Direct Preference Optimization (DPO). Applying our DPO-driven framework to the gemma3-1b model yields a significant improvement in translation quality, elevating it's COMET score from 0.703 to 0.747 on the English to German task. The results demonstrate that DPO offers an efficient and stable pathway for enhancing pre-trained NMT models through preference-based post-training.

Safety mechanisms for large language models (LLMs) remain predominantly English-centric, creating systematic vulnerabilities in multilingual deployment. Prior work shows that translating malicious prompts into other languages can substantially increase jailbreak success rates, exposing a structural cross-lingual security gap. We investigate whether such attacks can be mitigated through language-agnostic semantic similarity without retraining or language-specific adaptation. Our approach compares multilingual query embeddings against a fixed English codebook of jailbreak prompts, operating as a training-free external guardrail for black-box LLMs. We conduct a systematic evaluation across four languages, two translation pipelines, four safety benchmarks, three embedding models, and three target LLMs (Qwen, Llama, GPT-3.5). Our results reveal two distinct regimes of cross-lingual transfer. On curated benchmarks containing canonical jailbreak templates, semantic similarity generalizes reliably across languages, achieving near-perfect separability (AUC up to 0.99) and substantial reductions in absolute attack success rates under strict low-false-positive constraints. However, under distribution shift - on behaviorally diverse and heterogeneous unsafe benchmarks - separability degrades markedly (AUC $\approx$ 0.60-0.70), and recall in the security-critical low-FPR regime drops across all embedding models.

Accurate nutrient estimation from unstructured recipe text is an important yet challenging problem in dietary monitoring, due to ambiguous ingredient terminology and highly variable quantity expressions. We systematically evaluate models spanning a wide range of representational capacity, from lexical matching methods (TF-IDF with Ridge Regression), to deep semantic encoders (DeBERTa-v3), to generative reasoning with large language models (LLMs). Under the strict tolerance criteria defined by EU Regulation 1169/2011, our empirical results reveal a clear trade-off between predictive accuracy and computational efficiency. The TF-IDF baseline achieves moderate nutrient estimation performance with near-instantaneous inference, whereas the DeBERTa-v3 encoder performs poorly under task-specific data scarcity. In contrast, few-shot LLM inference (e.g., Gemini 2.5 Flash) and a hybrid LLM refinement pipeline (TF-IDF combined with Gemini 2.5 Flash) deliver the highest validation accuracy across all nutrient categories. These improvements likely arise from the ability of LLMs to leverage pre-trained world knowledge to resolve ambiguous terminology and normalize non-standard units, which remain difficult for purely lexical approaches. However, these gains come at the cost of substantially higher inference latency, highlighting a practical deployment trade-off between real-time efficiency and nutritional precision in dietary monitoring systems.

Critical analyses of emotion recognition technology have raised ethical concerns around task validity and potential downstream impacts, urging researchers to ensure alignment between their stated motivations and practice. However, these discussions have not adequately influenced or drawn from research on speech emotion recognition (SER). We address this gap by conducting a systematic survey of SER research to uncover what stated motivations drive this work and if they align with the datasets and emotions studied. We find that while SER research identifies appealing goals, such as well-situated voice-activated systems or healthcare applications, commonly-used datasets do not reflect these proposed deployment contexts, thus presenting a gap between motivations and research practices. We argue that such gaps engender ethical concerns, and that SER research should reassert itself with concrete use-cases to prevent misinterpretations, misuse, and downstream harms.

Subliminal learning describes a student language model inheriting a behavioral bias by fine-tuning on seemingly innocuous data generated by a biased teacher model. Prior work has begun to characterize this phenomenon but leaves open questions about the scope of signals it can transfer, the mechanisms that explain it, and the precision with which a bias can be encoded by seemingly unrelated data. We tackle all three problems by introducing subliminal steering, a variant of subliminal learning in which the teacher's bias is implemented not via a system prompt, as in prior work, but through a steering vector trained to maximize the likelihood of a set of target samples. First, we show that subliminal steering transfers complex multi-word biases, whereas prior work focused on single-word preferences, demonstrating a large scope of subliminally transferrable signals. Second, we provide mechanistic evidence that subliminal learning transfers not only the target behavioral bias, but also the steering vector itself, localized to the layers at which the teacher was steered. Finally, we show that the bias is encoded with surprising precision. We train a new steering vector directly on the subliminally-laden dataset and find that it attains high cosine similarity with the original vector.

Chain-of-Thought (CoT) has been shown to empirically improve Transformers' performance, and theoretically increase their expressivity to Turing completeness. However, whether Transformers can learn to generalize to CoT traces longer than those seen during training is understudied. We use recent theoretical frameworks for Transformer length generalization and find that – under standard positional encodings and a finite alphabet – Transformers with CoT cannot solve problems beyond $TC^0$, i.e. the expressivity benefits do not hold under the stricter requirement of length-generalizable learnability. However, if we allow the vocabulary to grow with problem size, we attain a length-generalizable simulation of Turing machines where the CoT trace length is linear in the simulated runtime up to a constant. Our construction overcomes two core obstacles to reliable length generalization: repeated copying and last-occurrence retrieval. We assign each tape position a unique signpost token, and log only value changes to enable recovery of the current tape symbol through counts circumventing both barriers. Further, we empirically show that the use of such signpost tokens and value change encodings provide actionable guidance to improve length generalization on hard problems.

Creating interactive STEM courseware traditionally requires HTML/CSS/JavaScript expertise, leaving barriers for educators. While generative AI can produce HTML codes, existing tools generate static presentations rather than interactive simulations, struggle with long documents, and lack pedagogical accuracy mechanisms. Furthermore, full regeneration for modifications requires 200–600 seconds, disrupting creative flow. We present MAIC-UI, a zero-code authoring system that enables educators to create and rapidly edit interactive courseware from textbooks, PPTs, and PDFs. MAIC-UI employs: (1) structured knowledge analysis with multi-modal understanding to ensure pedagogical rigor; (2) a two-stage generate-verify-optimize pipeline separating content alignment from visual refinement; and (3) Click-to-Locate editing with Unified Diff-based incremental generation achieving sub-10-second iteration cycles. A controlled lab study with 40 participants shows MAIC-UI reduces editing iterations (4.9 vs. 7.0) and significantly improves learnability and controllability compared to direct Text-to-HTML generation. A three-month classroom deployment with 53 high school students demonstrates that MAIC-UI fosters learning agency and reduces outcome disparities – the pilot class achieved 9.21-point gains in STEM subjects compared to -2.32 points in control classes. Our code is available at https://github.com/THU-MAIC/MAIC-UI.