Scientific and clinical decision-making depends on evidence from the primary literature, but existing standards for representing that evidence (FHIR Evidence, ECO, SEPIO, and the GA4GH Genomic Knowledge Standards) are oriented toward clinical-trial workflows, evidence codes, or single-variant assertions, and do not capture the fine-grained, domain-specific structure of claims in basic and pre-clinical research. We introduce a semantic model for scientific evidence with three core classes, specialize it for genetics, align it structurally to FHIR Evidence with a SEPIO-anchored credibility decomposition, and attach a compact dimensional vocabulary whose conditional-activation rules are validated by a SHACL schema for the implemented constraints. Using clinical variant interpretation as the driving use case, we evaluate the model through a human-AI annotation pilot over six genetics papers, yielding 28 evidence items and 95 source-anchored assertions, with a workflow that keeps curator-authored reference annotations distinct from AI-drafted annotations. Treating the pilot as a feasibility study rather than a benchmark, we argue that the model is a useful increment toward trustworthy, AI-ready infrastructure for variant interpretation: a reference data model and validation schema for representing genetic evidence.
We present Hakken, a domain-agnostic prediction and explanation system performing knowledge prediction, i.e., growing scientific knowledge by establishing novel relationships, ones that are not limited to the deductive hull of previous knowledge. Hakken uses a transformer-based prediction model built on temporal sequences of knowledge graphs extracted from vast bodies of research publications, fused with an LLM's semantic knowledge, to predict the presence and define the type of as-yet undocumented relationships between scientific concepts. It then calls a model-agnostic explanation framework to provide accompanying information for each prediction that allows scientists to evaluate the suggested new relationship. While general purpose, we demonstrate Hakken's practical capabilities by applying it to the biomedical domain. There, Hakken's prediction model establishes a new benchmark for time-aware multi-label relation prediction, and we show that the model's output stays coherent and informative over extended time spans in historic data. In addition, we scored 1.5 million above-confidence-threshold hypotheses related to aging, qualitatively validated batches of these predictions with biologists and progressed three of them for empirical validation in wet-lab. Two predictions with potentially significant impact in the context of drug discovery and repurposing were confirmed, introducing previously undocumented interactions between TP53 and BAMBI, and between RAF1 and TNF, to biomedical science.
Tarek R. Besold, Uchenna Akujuobi, Pablo Sanchez et al.· 0 citations
Pause-token methods improve LLM reasoning by inserting special tokens into sequences. Prior work explains these gains through computational expressivity. However, there is relatively little investigation into the training dynamics of pause tokens. We explore how pause tokens reshape the training dynamics of fine-tuning. Two controlled pilots expose distinct asymmetries. On a synthetic continual-learning task, masked pauses overwrite a previously-learned distribution roughly 4x less at matched final adaptation (H1, mode retention); on a synthetic math-reasoning probe, the boundary-adjacent token comes to encode substantially more downstream-step information (H2, non-myopic compression). We formalize a training rule consistent with both - Masked Boundary Pause (MBP), pause tokens placed at reasoning-step boundaries with their loss masked. Across 1B-8B Qwen and Llama models, MBP consistently improves reasoning, achieving gains of up to 6 points on math and 2.5 points on code, while preserving general language understanding abilities. We further demonstrate that this mode-preserving strategy extend gains to GRPO. These results recast pause tokens as a training-dynamics intervention on the retention-adaptation trade-off, rather than merely an inference-time computation device.
Jaehyeon Kim, Suhwan Kim, Nakyung Lee et al.· 0 citations
We evaluate three open-weight LLMs (Gemma3-12B from the USA, Bielik-11B-v3 from Poland, and Qwen3-4B from China) against World Values Survey Wave 7 data for 63 demographic personas across three countries, using normalized Wasserstein distance to quantify distributional misalignment. Contrary to expectations, no model favors its home country: the Chinese-built Qwen3-4B performs worst on its own Chinese population (W1 = 0.436, the highest misalignment in the entire model x country matrix). Targeted LoRA fine-tuning on the five worst-case personas, requiring fewer than 1,200 training pairs and under 15 minutes on a single GPU, reduces bias by 16.8% for Bielik-11B (p_Bonf = 0.002, d = -4.4) with all five targets improving. However, country-level decomposition reveals that fine-tuning redistributes rather than removes bias: Bielik's worst-case personas swap entirely from American to Chinese elderly, with zero overlap between pre- and post-correction sets. To our knowledge, this is the first study to target worst-case demographic personas with LoRA fine-tuning for cross-cultural bias mitigation.
Antoni Czolgowski, Abel Iyasele· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
This paper investigates structural priming in language model (LM) production, examining how preceding structural context influences sentence completion. While prior work has demonstrated priming effects in comprehension of structural alternations, it remained unclear whether these persist in production, where, when generating, an LM samples from many possible continuations at each step. We address this question through a series of controlled sentence-completion experiments on dative constructions. In line with prior work, we find that LMs are susceptible to structural priming, particularly in sentences that are semantically coherent. In terms of priming magnitude, we find that while there is a greater relative increase of double-object datives against our baselines, in line with inverse frequency effects, there is a larger absolute increase in prepositional-objects, the more frequently produced construction. Finally, we not only observe that structural priming is boosted by lexico-semantic coherence, but that structurally primed completions display greater levels of lexico-semantic repetition. Taken together, our evidence supports the view that structural priming in LMs operates across multiple levels of linguistic representation, facilitating, and facilitated by syntactic, lexical, and semantic alignment. Code: https://github.com/the-context-lab/primedproduction.
In many forms of reasoning, including arithmetic reasoning, generalizing across superficial changes in input format is effortless for humans: anyone who can solve 2+5 can also solve 'two plus five'. In contrast, LLMs are more brittle to surface variations of the prompts: for example, they solve numeric arithmetic problems almost perfectly but are substantially less accurate on verbal renditions of the same problems. Here, we ask whether generalization across formats can be predicted from the models' internals. Using attribution patching, we first independently localize the circuit that each model recruits to solve numeric arithmetic problems (2+5) vs. verbal ones, in three languages: English ('two plus five'), Spanish ('dos m\'as cinco'), and Italian ('due pi\`u cinque'); then, we test whether overlap with the model's own numeric circuit predicts its generalization to the verbal formats. Indeed, we find support for this idea at three levels: circuit overlap accounts for the relative difficulty of the three verbal formats, for which models generalize best, and for which items are solved correctly, rivaling supervised probes while requiring no labeled data.
Andrea Gregor de Varda, Sana Pandey, Pengrui Han et al.· 0 citations
Foundation models are beginning to reshape brain-signal analysis by moving the field beyond task-specific decoding pipelines toward reusable models pretrained on broad neural datasets. Magnetoencephalography (MEG) is a compelling but still underdeveloped target for this shift: it captures human cortical dynamics at millisecond resolution while offering stronger spatial interpretability than EEG, making it especially valuable for source-resolved studies of perception, language, cognition, and clinical brain function. Yet MEG foundation models remain at an early stage, with only a small number of MEG-specific and MEG-inclusive multi-modal models, modest pretraining corpora, and emerging but still limited benchmarks. This perspective lays down the basic concepts needed to understand MEG foundation models and provides a didactic overview of the field's key design choices, including tokenization, sensor- versus source-space representations, sensor-geometry encoding, backbone architectures, self-supervised objectives, and pretraining data. We then offer a roadmap for future development, organized around native MEG pretraining, adaptation of EEG foundation models, transfer from generic time-series models, and multi-modal integration with EEG, fMRI, MRI, behaviour, and stimulus features. We highlight the need for coordinated infrastructure, including diverse and reusable MEG datasets, rigorous evaluation across subjects, sites, tasks, and clinical settings, and responsible data-sharing practices that address consent, privacy, access, and governance.
Philipp Th\"olke, Hamza Abdelhedi, Yorguin Mantilla-Ramos et al.· 0 citations
Expert pruning reduces the memory and serving cost of Mixture-of-Experts (MoE) models by removing low-importance experts identified by the router, assuming router probabilities provide a reliable importance signal. We observe that this assumption breaks down under over-dispersed routing, a regime associated with aggressive load-balancing during training, in which tokens are distributed nearly uniformly across experts and importance signals collapse. In this regime, perplexity does not predict downstream task accuracy: on gpt-oss-20B, the lowest-perplexity pruning configuration yields the worst mathematical reasoning, while the highest-perplexity configuration preserves it. This does not occur under standard routing (e.g., Mixtral-8x7B-Instruct), where perplexity and accuracy degrade together. Pruning under over-dispersed routing also exposes a capability trade-off in which no single scoring metric dominates: activation-aware scoring preserves mathematical reasoning but severely degrades knowledge-intensive science (an 18-point gap on GPQA), whereas frequency-based scoring exhibits the reverse. We propose Minimax Expert Score Allocation (MESA), a domain-aware method that iteratively boosts importance scores for experts serving whichever domain is currently worst-affected, minimizing worst-case domain degradation rather than average accuracy. At 25% expert pruning MESA achieves the smallest worst-case degradation across domains, outperforming activation-aware baselines on 7 of 11 benchmarks at a correspondingly reduced memory footprint, and it generalizes to gpt-oss-120B, Gemma-4-26B-A4B, and OLMoE-1B-7B. Our results indicate that over-dispersed routing is a qualitatively distinct pruning regime in which standard assumptions fail, and that recognizing it is a prerequisite for principled expert pruning of load-balanced MoE models.
Berkcan Kapusuzoglu, Connor Pryor, Sangwoo Cho et al.· 0 citations
Large language models deployed in high-stakes settings frequently generate plausible but ungrounded claims. Standard retrieval-augmented generation (RAG) pipelines offer limited remedy, since they retrieve isolated passages without tracking cross-document evidence relationships or quantifying uncertainty. We introduce GRACE (Graph-grounded Reflective Agent Copilot Engine), a framework that deconstructs LLM responses into atomic claims and grounds them against trusted knowledge priors within a weighted bipartite graph. Edge weights encode the closeness of each claim to the priors, enabling weighted centrality analysis that classifies claims as Grounded, Refuted, or Boundary. Such classification identifies not just hallucinations but also novel or contested claims at the frontier of the model's knowledge. To efficiently allocate human or agent resources, we formulate a Return on Attention (RoA) objective that defers a claim to expert review only when its priority-weighted uncertainty exceeds the cost of verification. Claims verified by experts are promoted to new evidence anchors, closing a validator-LLM evolutionary loop that expands the knowledge base across iterations. We evaluate GRACE across multiple language models and on datasets spanning both general and domain-specific knowledge. Our results show that our knowledge base serves as a reliable foundation for retrieval that outperforms RAG baselines, and that the RoA framework efficiently selects valuable boundary knowledge for expert verification. These findings demonstrate that graph-structured representations combined with expert-in-the-loop verification can mitigate hallucination at the system level rather than at the generation level. Code available at https://github.com/johnsk95/grace_code
John Seon Keun Yi, Joshua R. Minot, Dokyun Lee· 0 citations
Learning rich medical concept representations is essential for EHR prediction. Text-attributed knowledge graphs (TKGs) provide a natural foundation by organizing heterogeneous medical relations together with textual semantics. However, most existing encoders process concepts uniformly across patients, despite the fact that a code's meaning and predictive value depend on patient-specific clinical context and trajectory. Learning patient-personalized concept representations from TKGs introduces two key challenges: (1) deciding how much KG context to incorporate for each observed code, and (2) aligning semantic information with the patient-specific relational structure. We propose REFINE, a KG-aware budgeted LLM graph refinement framework for patient-personalized medical concept encoding. Starting from a global TKG, REFINE constructs patient-specific temporal graphs. A sequential reinforcement learning policy selects a personalized KG expansion budget for each observed code. The resulting patient graph is processed by a heterogeneous GNN to capture relation-aware structural dependencies, while a frozen LLM uses graph-aware soft prompts to semantically refine concept representations. Experiments on MIMIC-III and MIMIC-IV show that REFINE consistently improves diverse EHR backbones, outperforms strong baselines, and demonstrates robust gains across component ablation, KG selection, and data insufficiency.
Mohsen Nayebi Kerdabadi, Arya Hadizadeh Moghaddam, Dongjie Wang et al.· 0 citations
Multilingual language models often produce inconsistent answers to semantically equivalent questions across languages, motivating methods to improve cross-lingual consistency (CLC). However, existing methods are typically evaluated using different models, tasks, and protocols, leaving their relative strengths unclear. In this work, we present a unified evaluation of representative CLC-enhancement methods for question answering, spanning inference-time interventions and post-training approaches across three model families and three closed-form benchmarks. The results show that post-training methods are generally more reliable, with direct distribution alignment consistently improving CLC across all model-dataset combinations, while other methods are more sensitive to answer format and the breadth of language coverage. Notably, cross-domain transfer is limited unless source and target tasks share similar output formats. We further investigate whether CLC enhancement hurts models' ability to respond differently *when needed*, that is, when asked culture-dependent questions. Across two benchmarks of culturally diverse question answering, we find no systematic degradation in controlled closed-form evaluation, whereas open-ended generation reveals occasional accuracy reductions, particularly for non-English responses. Our work highlights the need to evaluate CLC enhancement for both cross-domain robustness and culturally appropriate variation, informing future work in post-training and benchmark development.
Jirui Qi, Mingyang Wang, Hinrich Sch\"utze et al.· 0 citations
Chinese online comments often convey social meaning through indirect and playful language that is hard to interpret without context. Existing evaluations largely organize items around predefined phenomena or controlled pragmatic categories, leaving open whether models can distinguish plausible readings of what a naturally occurring comment is doing in a particular exchange. We introduce a benchmark for evaluating whether LLMs can recover such situated pragmatic meanings. From more than 200,000 public Chinese social media interaction records, we construct 4,735 human-validated diagnostic items, each pairing a target comment with reconstructed preceding context and plausible misreadings. We evaluate eight LLMs as both question writers and solvers in a cross-writer setting. The task is challenging: the strongest model achieves 81.42% leave-writer-out accuracy. Across all eight models, the mean leave-writer-out accuracy is 68.70% while human accuracy was 90.8%. Case analysis shows that models often recognize broad irony or playfulness while misidentifying the mechanism or interactional move.
Shiwei Hong, Junjie Ma, Emma Jiren Wang et al.· 0 citations
