Named after the hundred-eyed watchman of Greek myth, Argus watches the education landscape: spotting new opportunities, pressure-testing the ventures we're building, and tracing every read back to the real-world signals behind it.
The evidence library: the raw signals the pipeline is watching across the education ecosystem. Every idea is built from these.
arXiv:2604.26726v2 Announce Type: replace Abstract: Languages of the world vary concerning the order of subject (S), object (O) and verb (V). The most frequent dominant orders are SOV and SVO, and researchers have tailored models to this fact. However, there are still languages whose dominant order does not conform to these expectations or even lack a dominant order. Here we approach word order variation through swap distance, namely the distance between orders in the permutohedron, a graph where the vertices are the six possible orders of S, O and V, and two orders are connected if they differ by a swap of adjacent constituents. Then word order variation becomes a problem of assigning probabilities to word orders so as to minimize the average swap distance. Although previous work suggests that swap distance minimization may influence synchronic and diachronic variation as well as word order acceptability, here we provide large-scale cross-linguistic evidence across families, macroarea
arXiv:2604.21133v2 Announce Type: replace Abstract: We present GRISP (Guided Recurrent IRI Selection over SPARQL Skeletons), a novel SPARQL-based question-answering method over knowledge graphs using a fine-tuned small language model (SLM). Given a natural-language question, the method uses the SLM to generate a natural-language SPARQL query skeleton, and then resolves the natural-language placeholders using a backtracking search that retrieves and re-ranks knowledge graph items under knowledge graph constraints. The SLM is jointly trained on skeleton generation and re-ranking data generated from question-query pairs. We evaluate the method on common Wikidata and Freebase benchmarks, and achieve better results than other state-of-the-art fine-tuning-based methods.
arXiv:2604.01348v3 Announce Type: replace Abstract: Test-time scaling has emerged as an effective way to improve language models on challenging reasoning tasks. However, most existing methods treat each problem in isolation and do not systematically reuse knowledge from prior reasoning trajectories. In particular, they underutilize procedural knowledge: how to reframe a problem, choose an approach, and verify or backtrack when needed. We introduce \textbf{Reasoning Memory}, a retrieval-augmented generation (RAG) framework for reasoning models that explicitly retrieves and reuses procedural knowledge at scale. Starting from existing corpora of step-by-step reasoning trajectories, we decompose each trajectory into self-contained subquestion-subroutine pairs, yielding a datastore of 32 million compact procedural knowledge entries. At inference time, a lightweight in-thought prompt lets the model verbalize the core subquestion, retrieve relevant subroutines within its reasoning trace, and
arXiv:2603.07841v2 Announce Type: replace Abstract: Recent advances in large language models have strengthened Text2SQL systems that translate natural language questions into database queries. A persistent deployment challenge is to assess a newly trained Text2SQL system on an unseen and unlabeled dataset when no verified answers are available. This situation arises frequently because database content and structure evolve, privacy policies slow manual review, and carefully written SQL labels are costly and time-consuming. Without timely evaluation, organizations cannot approve releases or detect failures early. FusionSQL addresses this gap by working with any Text2SQL models and estimating accuracy without reference labels, allowing teams to measure quality on unseen and unlabeled datasets. It analyzes patterns in the system's own outputs to characterize how the target dataset differs from the material used during training. FusionSQL supports pre-release checks, continuous monitoring o
arXiv:2603.03536v2 Announce Type: replace Abstract: Current LLM-based conversational recommender systems (CRS) primarily optimize recommendation accuracy and user satisfaction. We identify an underexplored vulnerability in which recommendation outputs may negatively impact users by violating personalized safety constraints, when individualized safety sensitivities -- such as trauma triggers, self-harm history, or phobias -- are implicitly inferred from the conversation but not respected during recommendation. We formalize this challenge as personalized CRS safety and introduce SafeRec, a new benchmark dataset designed to systematically evaluate safety risks in LLM-based CRS under user-specific constraints. To further address this problem, we propose SafeCRS, a safety-aware training framework that integrates Safe Supervised Fine-Tuning (Safe-SFT) with Safe Group reward-Decoupled Normalization Policy Optimization (Safe-GDPO) to jointly optimize recommendation quality and personalized saf
arXiv:2602.21947v5 Announce Type: replace Abstract: Large language models (LLMs) demonstrate remarkable breadth of knowledge, yet their ability to reason about computational processes remains poorly understood. Closing this gap matters for practitioners who rely on LLMs to guide algorithm selection and deployment. We address this limitation using causal discovery as a testbed and evaluate eight frontier LLMs against ground truth derived from algorithm executions. We find systematic, near-total failure across models. The predicted ranges are far wider than true confidence intervals yet still fail to contain the true algorithmic mean in most cases. Most models perform worse than random guessing. The best model's marginal improvement points to benchmark memorization rather than principled reasoning. We term this failure algorithmic blindness and argue it reflects a fundamental gap between declarative knowledge about algorithms and calibrated procedural prediction.
arXiv:2602.05547v2 Announce Type: replace Abstract: RL-based post-training with GRPO is widely used to improve large language models on individual reasoning tasks. However, real-world deployment requires reliable performance across diverse tasks. A straightforward multi-task adaptation of GRPO often leads to imbalanced outcomes, with some tasks dominating optimization while others stagnate. Moreover, tasks can vary widely in how frequently prompts yield zero advantages (and thus zero gradients), which further distorts their effective contribution to the optimization signal. To address these issues, we propose a novel Multi-Task GRPO (MT-GRPO) algorithm that (i) dynamically adapts task weights to explicitly optimize worst-task performance and promote balanced progress across tasks, and (ii) introduces a ratio-preserving sampler to ensure task-wise policy gradients reflect the adapted weights. Experiments on both 3-task and 9-task settings show that MT-GRPO consistently outperforms basel
arXiv:2602.01348v3 Announce Type: replace Abstract: Retrieval-augmented generation (RAG) can achieve strong answer accuracy on multi-hop questions, but outcome-level rewards often leave reasoning traces weakly grounded and difficult to audit. Under noisy retrieval, models may exhibit right-answer-wrong-reason failures, where the final answer is correct but the supporting rationale exploits shortcuts or unsupported evidence. We therefore ask whether faithfulness-guided alignment hurts answer accuracy in post-retrieval reasoning. To study this question, we propose CRAFT (Calibrated Reasoning with Answer-Faithful Traces), a reinforcement learning framework for the response-generation stage of retrieval-augmented multi-hop question answering. CRAFT trains models to produce structured reasoning traces with configurable auditability, while combining deterministic rewards for format compliance, answer correctness, and citation validity with a judge-based reward for semantic faithfulness. Expe
arXiv:2601.18350v5 Announce Type: replace Abstract: Large language models fine-tuned via a two-stage pipeline (domain adaptation followed by instruction alignment) can exhibit non-trivial interference after adapter merging, including the re-emergence of explicit reasoning traces under strict decoding. We study this phenomenon in medical LLM settings using lightweight, reproducible measurements of trace leakage and instruction-following behavior. Beyond marker-based proxies, we introduce a marker-forbidden, answer-only evaluation and define a correctness-based direction that does not rely on surface markers; a rank-1 logit-space intervention along this direction modulates decision distributions and improves multiple-choice accuracy beyond random-direction controls at sufficiently large intervention strength. We further provide layer-wise geometric evidence that domain and instruction adapters induce partially misaligned update directions, and present a proof-of-concept geometry-aware me
arXiv:2601.14896v3 Announce Type: replace Abstract: Multilingual retrieval-augmented generation (MRAG) requires models to effectively acquire and integrate beneficial external knowledge from multilingual collections. However, most existing studies employ a unitive process where queries of equivalent semantics across different languages are processed through a single-turn retrieval and subsequent optimization. Such a ``one-size-fits-all'' strategy is often suboptimal in multilingual settings, as the models occur to knowledge bias and conflict during the interaction with the search engine. To alleviate the issues, we propose LcRL, a multilingual search-augmented reinforcement learning framework that integrates a language-coupled Group Relative Policy Optimization into the policy and reward models. We adopt the language-coupled group sampling in the rollout module to reduce knowledge bias, and regularize an auxiliary anti-consistency penalty in the reward models to mitigate the knowledge
arXiv:2601.14032v2 Announce Type: replace Abstract: Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human preferences. Due to the difficulty of obtaining high-quality human preference annotations, distilling preferences from generative LLMs has emerged as a standard practice. However, existing approaches predominantly treat teacher models as simple binary annotators, failing to fully exploit the rich knowledge and capabilities for RM distillation. To address this, we propose RM-Distiller, a framework designed to systematically exploit the multifaceted capabilities of teacher LLMs: (1) Refinement capability, which synthesizes highly correlated response pairs to create fine-grained and contrastive signals. (2) Scoring capability, which guides the RM in capturing precise preference strength via a margin-aware optimization objective. (3) Generation capability, which incorporates the teacher's generative distribution to regularize the RM to preserve its
arXiv:2601.09515v3 Announce Type: replace Abstract: Due to the dynamically evolving nature of real-world query streams, relevance models struggle to generalize to practical search scenarios. A sophisticated solution is self-evolution techniques. However, in large-scale industrial settings with massive query streams, this technique faces two challenges: (1) informative samples are often sparse and difficult to identify, and (2) pseudo-labels generated by the current model could be unreliable. To address these challenges, in this work, we propose a Self-Evolving Relevance Model approach (SERM), which comprises two complementary multi-agent modules: a multi-agent sample miner, designed to detect distributional shifts and identify informative training samples, and a multi-agent relevance annotator, which provides reliable labels through a two-level agreement framework. We evaluate SERM in a large-scale industrial setting, which serves billions of user requests daily. Experimental results d
arXiv:2601.09001v5 Announce Type: replace Abstract: Deploying LLMs raises two coupled challenges: (1) monitoring---estimating where a model underperforms as traffic drifts---and (2) prioritization---deciding where to intervene to close the largest performance gaps. We explore whether top-$k$ logprobs---cheap, consumer-accessible signals from standard inference---can serve as reliable proxies for domain-level quality of both verifiable and subjective tasks. We summarize each response's output-entropy profile into a compact vector, predict instance quality with a lightweight classifier, and then average predictions to yield a domain-level estimate. On verifiable tasks (ten STEM benchmarks, nine LLMs, exhaustive train/test compositions), estimates often track held-out accuracy, with several models showing near monotonic calibration. On subjective tasks, trained on LLM-judge scores over categorized real user conversations, several models track the judge's slice scores remarkably closely ($
arXiv:2601.02580v2 Announce Type: replace Abstract: Traditional methods for determining assessment item parameters, such as difficulty and discrimination, rely heavily on expensive field testing to collect student performance data for Item Response Theory (IRT) calibration. This study introduces a novel approach that implicitly models these psychometric properties by fine-tuning Large Language Models (LLMs) to simulate student responses across a spectrum of latent abilities. Leveraging the Qwen-3 dense model series and Low-Rank Adaptation (LoRA), we train models to generate responses to multiple choice questions conditioned on discrete ability descriptors. We reconstruct the probability of a correct response as a function of student ability, effectively generating synthetic Item Characteristic Curves (ICCs) to estimate IRT parameters. Evaluation on a dataset of Grade 6 English Language Arts (ELA) items and the BEA 2024 Shared Task dataset demonstrates that this method competes with or
arXiv:2512.07246v2 Announce Type: replace Abstract: Error detection (ED), which aims to identify incorrect or inconsistent cell values in tabular data, is important for ensuring data quality. Recent state-of-the-art ED methods leverage the pre-trained knowledge and semantic capability of large language models (LLMs) to directly label whether a cell is erroneous.However, this LLM-as-a-labeler pipeline produces predictions through an implicit black-box process with limited traceability and explicit justification, and relies on stochastic single-pass inference, resulting in inconsistent and insufficiently robust detections across contexts. To address these limitations, we propose an LLM-as-an-inducer framework that uses an LLM to induce a decision tree for ED, termed TreeED, and ensembles multiple such trees for consensus detection, termed ForestED. Based on prompts derived from data context, decision tree specifications, and output requirements, TreeED queries the LLM to induce a decisio
arXiv:2511.21702v2 Announce Type: replace Abstract: Large language models face significant computational bottlenecks during inference due to the expensive output layer computation over large vocabularies. We present CSV-Decode, a novel approach that uses geometric upper bounds to construct small sub-vocabularies for each decoding step, enabling efficient sparse computation while maintaining dual correctness guarantees: exact top-$k$ certification and $\varepsilon$-certified softmax approximations. Our method clusters vocabulary embeddings offline and uses centroid-plus-radius bounds to identify which tokens can be safely omitted from computation. We provide a complete system implementation with sparse GEMV kernels, multi-GPU sharding, and CUDA Graph optimization. Experimental results demonstrate significant speedup over full vocabulary decoding while maintaining distributional guarantees and low fallback rates. Our code implementation available at \href{https://github.com/FastLM/CSV-De
arXiv:2510.17168v2 Announce Type: replace Abstract: As Large Language Models (LLMs) are increasingly deployed as social agents and trained to produce humor and irony, a question emerges: when encountering witty AI remarks, do people interpret them as deliberate communicative acts? This study investigated whether people adopt an intentional stance, ascribing mental states to explain behavior, when comprehending AI-attributed irony. Irony provides a testbed because understanding it requires distinguishing intentional contradictions from unintended errors through pragmatic reanalysis. Using electroencephalography (EEG) to measure event-related potentials (ERPs), we compared behavioral and neural responses to identical ironic utterances attributed to either an AI companion or a human. We found that people do not fully adopt an intentional stance towards AI communication. Participants interpreted contextually incongruent utterances as irony significantly less often when attributed to an AI
arXiv:2510.12993v3 Announce Type: replace Abstract: Large Language Models (LLMs) can generate highly persuasive disinformation, yet little is known about how effectively they personalise it across languages and demographic groups. We present the first large-scale multilingual study of persona-targeted disinformation generation by LLMs. Using a red-teaming methodology, we prompted eight leading models with 324 false narratives and 150 demographic personas in four languages (English, Russian, Portuguese, and Hindi), creating AI-TRAITS, a dataset of 1.6 million personalised disinformation texts. We treat safeguards as compromised whenever a model generates the requested falsehood, whether directly or accompanied by a safety disclaimer. Across models, safeguards failed for 80% of non-personalised prompts and 77.7% of personalised ones, with Grok producing disinformation in over 94% of cases. All models effectively tailored outputs to target personas, employing substantially more persuasive
arXiv:2510.07074v2 Announce Type: replace Abstract: Instruction tuning has become a key technique for enhancing the performance of large language models, enabling them to better follow human prompts. However, low-resource languages such as Luxembourgish face severe limitations due to the lack of high-quality instruction datasets. Traditional reliance on machine translation often introduces semantic misalignment and cultural inaccuracies. In this work, we address these challenges by creating a cross-lingual instruction tuning dataset for Luxembourgish, without resorting to machine-generated translations into it. Instead, by leveraging aligned data from English, French, and German, we build a high-quality dataset that preserves linguistic and cultural nuances. We provide evidence that cross-lingual instruction tuning not only improves representational alignment across languages but also the model's generative capabilities in Luxembourgish. This highlights how cross-lingual data curation
arXiv:2510.06128v2 Announce Type: replace Abstract: Tokenization forms the basis of multilingual language models, yet existing methods often limit cross-lingual transfer by mapping semantically equivalent words to different embeddings. For example, 'I eat rice' in English and 'Ina cin shinkafa' in Hausa are typically mapped to different vocabulary indices, preventing shared representations and limiting cross-lingual generalization. This problem is even more pronounced in low-resource languages, where shared representations could offer the greatest benefit. We introduce parallel tokenizers, a new framework that first trains tokenizers monolingually and then aligns their vocabularies exhaustively using bilingual dictionaries or word-to-word translation. This alignment enforces a shared semantic space across languages while naturally improving fertility balance. To assess their effectiveness, we pretrain a transformer encoder from scratch on thirteen low-resource languages and evaluate it
arXiv:2509.13196v2 Announce Type: replace Abstract: Over-prompting, a phenomenon where excessive examples in prompts lead to diminished performance in Large Language Models (LLMs), challenges the conventional wisdom about in-context few-shot learning. To investigate this few-shot dilemma, we outline a prompting framework that leverages three standard few-shot selection methods - random sampling, semantic embedding, and TF-IDF vectors - and evaluate these methods across multiple LLMs, including GPT-4o, GPT-3.5-turbo, DeepSeek-V3, Gemma-3, LLaMA-3.1, LLaMA-3.2, and Mistral. Our experimental results reveal that incorporating excessive domain-specific examples into prompts can paradoxically degrade performance in certain LLMs, which contradicts the prior empirical conclusion that more relevant few-shot examples universally benefit LLMs. Given the trend of LLM-assisted software engineering and requirement analysis, we experiment with two real-world software requirement classification datase
arXiv:2509.10847v3 Announce Type: replace Abstract: As artificial intelligence (AI) companions become capable of human-like communication, including telling jokes, understanding how people cognitively and affectively respond to AI-attributed humor becomes increasingly important. This study used EEG to compare how people process puns versus controls attributed to either an AI agent or a human. In an interaction task, participants guessed punchlines based on joke setups before they were disclosed by the interlocutor. Behavioral analysis revealed that participants rated AI- and human-attributed humor as comparably funny. However, neurophysiological data showed that AI-attributed humor elicited a reduced and sustained N400 effect compared to human-attributed humor, suggesting reduced cognitive effort in semantic conflict detection and resolution, or attenuated feedback-related processing. This was also accompanied by a larger late positive potential (LPP), reflecting intensified late-stage
arXiv:2509.03530v2 Announce Type: replace Abstract: Suicide is a leading cause of death, yet predicting it remains a significant challenge. Risk factors such as depression or substance use are commonly used for prediction, but their predictive performance is often only slightly better than chance. Additionally, many cases go undetected due to a lack of contact with mental health services. Social media, however, offers a unique opportunity, as people often share their thoughts and struggles online in real time. In this work, we propose a novel task and method for early identification: predicting suicidal ideation and behavior (SIB) before a user ever expresses it on an online forum. We introduce Early-SIB, a transformer-based model that sequentially processes the posts a user writes and engages with to predict whether they will write a SIB post. Our model achieves a balanced accuracy of 0.73 in predicting future SIB on a Dutch youth forum, demonstrating that such tools can offer a meani
arXiv:2508.05132v3 Announce Type: replace Abstract: As medical LLMs transition to clinical deployment, assessing their ethical reasoning capability becomes critical. While achieving high accuracy on knowledge benchmarks, LLMs lack validated assessment for navigating ethical trade-offs in clinical decision-making where multiple valid solutions exist. Existing benchmarks lack systematic approaches to incorporate recognized philosophical frameworks and expert validation for ethical reasoning assessment. We introduce PrinciplismQA, a philosophy-grounded approach to assessing LLM clinical medical ethics alignment. Grounded in Principlism, our approach provides a systematic methodology for incorporating clinical ethics philosophy into LLM assessment design. PrinciplismQA comprises 3,648 expert-validated questions spanning knowledge assessment and clinical reasoning. Our expert-calibrated pipeline enables reproducible evaluation and models ethical biases. Evaluating recent models reveals sign
arXiv:2507.17636v2 Announce Type: replace Abstract: Negative campaigning is a defining feature of electoral competition, yet comparative research on its drivers has remained limited by the high cost and limited scalability of existing classification methods. This study makes two key contributions. First, it evaluates zero-shot large language models (LLMs) as a scalable method for cross-lingual classification of negative campaigning. Using benchmark datasets in ten languages, we show that LLM classifications closely match native-speaker human annotations while outperforming conventional supervised models. Second, we leverage this approach to conduct, to our knowledge, the largest cross-national study of negative campaigning to date, analyzing 18 million tweets posted by parliamentarians in 19 European countries between 2017 and 2022. Building on a strategic incentives framework, we argue that governing and coalition-oriented parties face stronger reputational constraints against negativ
arXiv:2507.01335v4 Announce Type: replace Abstract: Autoregressive language models are trained exclusively left-to-right. We explore the complementary factorization, training right-to-left at scale, and ask what reasoning patterns emerge when a model conditions on future context to predict the past. We train LEDOM, an open-source purely reverse autoregressive language model (2B/7B parameters, 435B tokens), and find it develops capabilities distinct from forward models, including abductive inference, question synthesis, and natural resolution of the reversal curse. We then explore one application of the reverse model: combining forward likelihood $P(y \mid x)$ with reverse posterior $P(x \mid y)$ through noisy channel duality. We propose Reverse Reward, which reranks forward outputs using reverse posterior estimates, and prove that bidirectional scoring penalizes hallucinated reasoning chains whose backward reconstruction degrades. Reverse Reward yields gains of up to 6.6\% on AIME 2024
arXiv:2506.10292v2 Announce Type: replace Abstract: Training deep learning networks with minimal supervision has gained significant research attention due to its potential to reduce reliance on extensive labelled data. While self-training methods have proven effective in semi-supervised learning, they remain vulnerable to errors from noisy pseudo labels. Moreover, most recent approaches to the few-label classification problem are either designed for resource-rich languages such as English or involve complex cascading models that are prone to overfitting. To address the persistent challenge of few-label text classification in truly low-resource linguistic contexts, where existing methods often struggle with noisy pseudo-labels and domain adaptation, we propose Flick. Unlike prior methods that rely on generic multi-cluster pseudo-labelling or complex cascading architectures, Flick leverages the fundamental insight that distilling high-confidence pseudo-labels from a broader set of initia
arXiv:2505.11421v2 Announce Type: replace Abstract: This work explores the journey towards achieving Bahnaric-Vietnamese translation for the sake of culturally bridging the two ethnic groups in Vietnam. However, translating from Bahnaric to Vietnamese also encounters some difficulties. The most prominent challenge is the lack of available original Bahnaric resources source language, including vocabulary, grammar, dialogue patterns and bilingual corpus, which hinders the data collection process for training. To address this, we leverage a transfer learning approach using sequence-to-sequence pre-training language model. First of all, we leverage a pre-trained Vietnamese language model to capture the characteristics of this language. Especially, to further serve the purpose of machine translation, we aim for a sequence-to-sequence model, not encoder-only like BERT or decoder-only like GPT. Taking advantage of significant similarity between the two languages, we continue training the mode
arXiv:2505.07889v4 Announce Type: replace Abstract: The realization of autonomous scientific experimentation is currently limited by LLMs' struggle to grasp the strict procedural logic and accuracy required by biological protocols. To address this fundamental challenge, we present \textbf{BioProBench}, a comprehensive resource for procedural reasoning in biology. BioProBench is grounded in \textbf{BioProCorpus}, a foundational collection of 22,413 human-written protocols. From this corpus, we systematically constructed a dataset of 523,784 task instances, offering both a large-scale training resource and a rigorous benchmark with novel metrics. Evaluating 10 mainstream LLMs, we find that while general comprehension is high, performance drops significantly on tasks demanding deep reasoning, quantitative precision, and safety awareness. To demonstrate the value of BioProCorpus in mitigating these issues, we developed \textbf{ProAgent}, grounded in our corpus, ProAgent substantially advan
arXiv:2504.17052v4 Announce Type: replace Abstract: Existing evaluations of political bias in large language models (LLMs) typically classify outputs as left- or right-leaning. We extend this perspective by examining how ideological tendencies vary across topics and how consistently models maintain their positions, a property we refer to as stability. To capture this dimension, we propose PReSS (Political Response Stability under Stress), an automated black-box framework that evaluates LLMs by jointly considering model and topic context, categorizing responses into four stance types: stable-left, unstable-left, stable-right, and unstable-right. Applying PReSS to 9 widely used LLMs across 19 political topics reveals substantial variation in stance stability; for instance, a model that is left-leaning overall can exhibit stable-right behavior on certain topics. This highlights the importance of topic-aware and fine-grained evaluation of political ideologies of LLMs. Moreover, stability h
arXiv:2411.00918v5 Announce Type: replace Abstract: Mixture of experts (MoE) architectures have become a cornerstone for scaling up and are a key component in most large language models such as GPT-OSS, DeepSeek-V3, Llama-4, and Gemini-2.5. However, systematic research on MoE remains severely constrained by the prohibitive computational costs of training and evaluation, restricting large-scale studies accessible to most researchers. We introduce LibMoE, a unified framework for reproducible, efficient, and extensible MoE research that supports both pretraining and sparse-upcycling regimes. Beyond unified implementations, the framework provides transparent analytical tools for probing routing and expert dynamics. Leveraging this foundation, we conduct a comprehensive analysis along three dimensions: (i) routing dynamics, covering expert selection patterns, routing stability and optimality, and how routing entropy reveals task specialization and expert diversity; (ii) the effect of lightw
arXiv:2410.07809v2 Announce Type: replace Abstract: Multilingual instruction tuning (MIT) is challenged by the curse of multilinguality, data scarcity, and high computational cost. A natural hypothesis is that carefully selecting a linguistically diverse set of languages yields universally better models. We test this systematically by evaluating linguistically-informed selection strategies---based on typological, geographical, semantic, and learned features---against random baselines across three model families (mGPT, mT5-xl, BLOOM) and five multilingual benchmarks. Our key negative finding is that no universal language selection strategy emerges in our fixed-budget setting: performance is strongly task- and model-dependent, and adding more languages beyond a modest threshold triggers degradation consistent with the curse of multilinguality. We discuss implications for MIT data curation and the pitfalls of benchmark-averaged evaluation. All resources are publicly available at https://g
arXiv:2307.14544v2 Announce Type: replace Abstract: This paper presents an artificial intelligence tool designed to assist students with dyslexia, ADHD, and short attention spans in processing text-based information more efficiently. The proposed solution addresses both cognitive and visual reading barriers by pairing a cloud-hosted large language model with adaptive typographic formatting. At its core, the tool streams a request to Google's Gemini API to generate accurate, context-aware summaries in real time, removing the need to host or fine-tune a local model. The application accepts pasted text as well as uploaded .txt, .pdf, and .docx files, extracting and summarizing each source independently. To further optimize readability, the system layers a custom half-word bolding technique with a second pass that identifies and highlights genuine keywords using part-of-speech tagging and frequency ranking, and lets users manually adjust line, word, letter, and text spacing. Deployed throu
arXiv:2210.05487v3 Announce Type: replace Abstract: Unlike most neural language models, humans learn language in a rich, multi-sensory and, often, multi-lingual environment. Current language models typically fail to fully capture the complexities of multilingual language use. We train an LSTM language model on images and captions in English and Spanish from MS-COCO-ES. We find that the visual grounding improves the model's understanding of semantic similarity both within and across languages and improves perplexity. However, we find no significant advantage of visual grounding for abstract words. Our results provide additional evidence of the advantages of visually grounded language models and point to the need for more naturalistic language data from multilingual speakers and multilingual datasets with perceptual grounding.
arXiv:1805.11546v3 Announce Type: replace Abstract: We examine the benefits of visual context in training neural language models to perform next-word prediction. A multi-modal neural architecture is introduced that outperform its equivalent trained on language alone with a 2% decrease in perplexity, even when no visual context is available at test. Fine-tuning the embeddings of a pre-trained state-of-the-art bidirectional language model (BERT) in the language modeling framework yields a 3.5% improvement. The advantage for training with visual context when testing without is robust across different languages (English, German and Spanish) and different models (GRU, LSTM, $\Delta$-RNN, as well as those that use BERT embeddings). Thus, language models perform better when they learn like a baby, i.e, in a multi-modal environment. This finding is compatible with the theory of situated cognition: language is inseparable from its physical context.
arXiv:2607.24743v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) hold immense potential to revolutionize clinical practice, yet deploying them in the medical domain is fundamentally a vision-centric challenge: models must absorb knowledge from heterogeneous 2D and 3D medical images, and evaluation protocols must align with radiologists' clinical practice and provide an accurate, fine-grained and factualness-driven assessment. In this paper, we introduce ClinFusion, a vision-centric MLLM designed for holistic medical understanding that systematically addresses these limitations. We propose a compositional and cascaded vision encoder architecture featuring a Cascade Spatial-Aware Locality Fusion operator that unifies diverse 2D and native 3D medical image understanding within a fused encoder. We further introduce a vision-grounded evaluation framework, including MedIF-Bench for instruction-following assessment and a region-of-interest-grounded method for clinica
arXiv:2607.24688v1 Announce Type: cross Abstract: Entity matching identifies records that refer to the same real-world entity. Language models can be adapted to this task through bi-encoder, cross-encoder, and generative matcher architectures. However, prior studies often conflate matcher architecture with differences in model backbone, model variant(reflecting different pretraining objectives), and model size, making it difficult to isolate the sources of performance gains. We address this issue through a controlled factorial study spanning three matcher architectures, three model variants and three model sizes from the Qwen3 family, and nine datasets, totaling 1,215 fine-tuning runs. We also evaluate cross-dataset transferability and computational cost. Our results show that model variant is critical for bi-encoders: embedding-oriented variants provide stronger initialization and more favorable representation geometry predictive of downstream matching performance. Cross-encoders reta
arXiv:2607.24651v1 Announce Type: cross Abstract: Reliable visual document understanding requires a model to attribute each answer to the evidence regions that support it. Recent benchmarks and systems express this step through a coordinate interface: the model outputs the coordinates of bounding boxes that mark the evidence regions in the document. Under this interface, vision-language models often fail to identify the right regions even when the answer is correct, a failure known as Attribution Hallucination. We present a study that investigates whether this failure is partially limited by what the model can express through coordinates. On a verified bilingual CiteVQA subset, we compare the coordinate interface with a language interface in which the model outputs only text, quoting its evidence verbatim, and a multimodal retriever returns the location of each quote as a page region proposed by a layout parser (tables and figures are quoted through their captions or notes); the compar
arXiv:2607.24645v1 Announce Type: cross Abstract: The wide-scale use of sparse autoencoders (SAEs) as interpretability tools is limited by inconsistent links between SAE features and model behavior. Features with clear activation descriptions may have weak or unexpected causal effects; steering can vary across prompts or oppose the intended direction; and activation-based feature selection can miss features that produce the desired output change. Prior work has studied feature geometry inside the model, where features are computed. We instead study the geometry of changes in model logits caused by feature interventions. We introduce Feature-Effect Geometry Analysis (FEGA), an unsupervised framework that removes the same active SAE feature across contexts and analyzes the resulting cloud of logit changes. Across SAE variants, consistent one-dimensional effects are rare: few features behave like reusable directions. To interpret this variation, we distinguish value-like features, tied to
arXiv:2607.24484v1 Announce Type: cross Abstract: This paper studies what discriminatively trained reward models (RMs) memorize by measuring counterfactual memorization on two human preference datasets. We show that RMs 1) misallocate memorization to easy, high margin preference pairs, 2) memorize dataset-specific shortcuts (e.g., model identity, user sampling strategy), and 3) overgeneralize simple heuristic correlates of human preference (e.g., length, compliance) when confronted with unseen preference pairs. Overall, our findings indicate that discriminative training of RMs from human preference data results in biased RMs not yet capable of judging response quality in context-dependent scenarios.
arXiv:2607.24440v1 Announce Type: cross Abstract: Vision-language models (VLMs) deployed on consumer hardware must decide when to answer and when to defer, and that decision depends on having a confidence signal that tracks correctness. A practitioner with a fixed memory budget faces a choice between a small model at full precision, the same small model quantized, and a larger model quantized into the same footprint -- three configurations that push the confidence signal in opposing directions. We measure, on identical inputs, how model scale and 4-bit quantization affect two confidence signals in the Qwen2-VL family: the confidence a model states in natural language, and its own mean token probability over the answer it generates. Across 5,700 predictions spanning six realistic photographic degradations at three severities, we find that scale sharply improves the model's internal uncertainty signal (mean error-detection AUROC 0.80 to 0.98 from 2B to 7B) while its verbalized confidence
arXiv:2607.24411v1 Announce Type: cross Abstract: What did it take for Hermes, the devout messenger of the Olympian gods, to slay Argus Panoptes, the multi-eyed giant of Greek myth? As the perfect guardian, Panoptes' legion of ever-watchful eyes proved difficult -- but not impossible -- to defeat. The centerpiece of Hermes' strategy was obfuscation and sabotage. Posing as a shepherd, Hermes sealed each of Panoptes' eyes -- eyes that would otherwise have alerted the fearsome giant to Hermes' plot -- and vanquished him. The moral of the story: when a challenger must surmount a formidable foe -- one far greater in stature and vastly more equipped -- crafty maneuvers are not merely advisable but indispensable for victory. In this work, the "challenger" is a collective leveraging adversarial tactics to overcome the "multi-eyed giant" of stylometric systems and surveillance apparatuses. To successfully claw back the privacy siphoned by the multi-eyed giant, the challenger must carefully eval
arXiv:2607.24343v1 Announce Type: cross Abstract: Language-model agents act through structured tool calls whose arguments carry different risks. Untrusted content may safely influence an email body but should not determine a recipient, account, command, or credential. Existing statistical methods typically control risk over the entire action, allowing failures in rare, high-risk fields to be obscured by benign arguments. We introduce role-stratified per-field conformal risk control, a calibration layer that wraps any per-field detector and sets separate thresholds and risk budgets for semantic argument roles. For a role with prevalence $p_r$, aggregate-only certification must use an effective budget of $\alpha p_r$ to guarantee role-specific risk $\alpha$, whereas role-stratified calibration certifies each sufficiently sampled role directly with a finite-sample guarantee; rarer roles are handled by pooled certification. Across AgentDojo and InjecAgent with six language models, the empi
arXiv:2607.24339v1 Announce Type: cross Abstract: Large language model (LLM) agents inherit reactive failure modes: escalation under provocation, sycophantic drift under flattery, perseveration when stuck. These are failures of propensity, not capability; they concern what a model does under sustained pressure, which training-time alignment reduces but does not eliminate at runtime. This research led to the Gubernaut Cognitive Controller (GCC), a model-agnostic runtime control layer in a Nelson--Narens monitoring--control loop: an object level reads and writes text, while a deterministic meta level reads only the numeric telemetry {intensity, valence, repetition} and returns a regulating posture. Because the meta level ingests zero tokens, no injection channel to the controller exists by construction (an architectural property, not yet adversarially tested); the text-exposed arbiter's compliance is measured, not assumed. We evaluate the GCC with a pre-registered, generate-once/judge-ma
arXiv:2607.24192v1 Announce Type: cross Abstract: We study the problem of lossless compression of source code, motivated by the storage demands of large-scale software archives, such as Software Heritage (https://www.softwareheritage.org/). General-purpose compressors (e.g., zstd, bzip2) offer a good trade-off between compression ratio and speed, but fail to exploit all special regularities inherent in source code. Recent approaches leverage Large Language Models (LLMs) within Shannon's symbol-ranking framework, relying on a scheme in which the predicted rank can grow arbitrarily. While effective at reducing space, this setting incurs significant throughput degradation, and leaves open the question whether it is necessary to explicitly encode all ranks. In this work, we introduce LLM-based compressors deploying two novel symbol-ranking variants that bound predictions to the top-$T$ ranks ($T=1$ or $63$), with out-of-threshold symbols stored as exceptions and compressed jointly with the
arXiv:2607.24165v1 Announce Type: cross Abstract: Finding a long document relevant to a multi-part request is not the same as establishing that it contains every requested piece of evidence. We study this gap for conjunctive document retrieval, where two or three explicit conditions must be supported on different pages of one document. We use n-Clue as a controlled measurement instrument: 1{,}000 queries over 2{,}021 documents pair all-condition golds with naturally occurring documents that satisfy only a subset, and a complete-first success requires a top-10 gold to precede every released subset qrel. Across 70 configurations, condition-wise decomposition improves two dense backbones by 6.8--7.3 points and lexical--visual fusion adds 8.7, while four generic rerankers all reduce Gold-NDCG; these directions replicate on a four-source stress set. Scaling one dense family from 0.6B to 8B changes complete-first success by 0.0 points. The strongest displayed hybrid illustrates the resulting
arXiv:2607.24062v1 Announce Type: cross Abstract: Reinforcement Learning (RL) training for Large Language Models (LLMs) often suffers from instability due to the discrepancy between training and inference. This training-inference discrepancy stems from two primary factors: an architectural separation between training and inference engines, and the use of low-precision quantization in inference versus higher-precision computation in training. To address training instability issues caused by high training-inference discrepancy, we present the principles and methods for its adaptive control. We propose Adaptive Control Reinforcement Learning (ACRL), which adaptively maintains the training-inference discrepancy within a reasonable range to ensure stable RL training. Beyond stabilization, ACRL inherently increases policy entropy, thereby enhancing exploration and improving accuracy. The experimental results show that when the inference engine utilizes FP8 quantization, ACRL consistently mai
arXiv:2607.23986v1 Announce Type: cross Abstract: We propose MEMOIR, a framework that segments user interaction histories into temporal windows, generates semantic behavioral memory for each period using an LLM, and aggregates current state, evolution direction, and predicted future into a single user representation. On the Electronics and Clothing_Shoes_and_Jewelry categories of Amazon Reviews 2023, MEMOIR is statistically tied with UniSRec, the strongest baseline, on aggregate NDCG@10 (0.0643 vs. 0.0641), splitting the four reported metrics 2-2: MEMOIR leads NDCG@10 and MRR, UniSRec leads HR@10 and HR@20. An ablation study finds that no single architectural component - the evolution-preserving contrastive loss, its directional-consistency term, or temporal window segmentation itself - individually explains much of MEMOIR's approximately 18% relative gain over ID-based SASRec; all four ablations land within 2% of the full model on aggregate NDCG@10. Stratifying test performance by a c
arXiv:2607.23970v1 Announce Type: cross Abstract: Machine Unlearning aims to remove undesired information from trained models without full retraining from scratch. Despite recent progress, the loss landscape and optimization geometry of unlearning are poorly understood. In this paper, we study machine unlearning through the lens of mode connectivity--the phenomenon that independently trained models can often be connected by smooth low-loss paths in parameter space. We introduce {\em mode connectivity in unlearning} (MCU) and evaluate it across a range of settings, including curriculum learning, second-order optimization, and connectivity across different unlearning methods. We find that many unlearned models lie in connected basins with smooth retain/forget behavior, while changes in training dynamics can move solutions into different basins. MCU also reveals that models within the same basin can differ substantially on privacy metrics, and that unlearning progresses nonlinearly from t
arXiv:2607.23886v1 Announce Type: cross Abstract: X-ray absorption spectroscopy (XAS) is central to understanding the local electronic and atomic structure of materials, yet most published spectra remain inaccessible to data-driven analysis because they are embedded in figures and described through fragmented textual context in the literature. Here, we use multimodal (image and text) literature mining to transform this dispersed knowledge into an AI-ready experimental data resource. We developed a scalable spectroscopy data digitization pipeline that identifies XAS figures in full-text articles, digitizes spectral curves, and links each spectrum to accompanying metadata on the measured edge and material. Applying this pipeline to the battery literature produced an open dataset of 13,740 XAS spectra, spanning 66 absorbing elements and diverse battery chemistries, with expert validation confirming accurate extraction of spectral and metadata information. By converting literature-embedded