Roi Reichart

Zorik Gekhman, Roee Aharoni, Eran Ofek et al. · deepmind

While reasoning in LLMs plays a natural role in math, code generation, and multi-hop factual questions, its effect on simple, single-hop factual questions remains unclear. Such questions do not require step-by-step logical decomposition, making the utility of reasoning highly counterintuitive. Nevertheless, we find that enabling reasoning substantially expands the capability boundary of the model's parametric knowledge recall, unlocking correct answers that are otherwise effectively unreachable. Why does reasoning aid parametric knowledge recall when there are no complex reasoning steps to be done? To answer this, we design a series of hypothesis-driven controlled experiments, and identify two key driving mechanisms: (1) a computational buffer effect, where the model uses the generated reasoning tokens to perform latent computation independent of their semantic content; and (2) factual priming, where generating topically related facts acts as a semantic bridge that facilitates correct answer retrieval. Importantly, this latter generative self-retrieval mechanism carries inherent risks: we demonstrate that hallucinating intermediate facts during reasoning increases the likelihood of hallucinations in the final answer. Finally, we show that our insights can be harnessed to directly improve model accuracy by prioritizing reasoning trajectories that contain hallucination-free factual statements.

Ariel Goldstein, Eric Ham, Mariano Schain et al. · princeton

Deep Language Models (DLMs) provide a novel computational paradigm for understanding the mechanisms of natural language processing in the human brain. Unlike traditional psycholinguistic models, DLMs use layered sequences of continuous numerical vectors to represent words and context, allowing a plethora of emerging applications such as human-like text generation. In this paper we show evidence that the layered hierarchy of DLMs may be used to model the temporal dynamics of language comprehension in the brain by demonstrating a strong correlation between DLM layer depth and the time at which layers are most predictive of the human brain. Our ability to temporally resolve individual layers benefits from our use of electrocorticography (ECoG) data, which has a much higher temporal resolution than noninvasive methods like fMRI. Using ECoG, we record neural activity from participants listening to a 30-minute narrative while also feeding the same narrative to a high-performing DLM (GPT2-XL). We then extract contextual embeddings from the different layers of the DLM and use linear encoding models to predict neural activity. We first focus on the Inferior Frontal Gyrus (IFG, or Broca's area) and then extend our model to track the increasing temporal receptive window along the linguistic processing hierarchy from auditory to syntactic and semantic areas. Our results reveal a connection between human language processing and DLMs, with the DLM's layer-by-layer accumulation of contextual information mirroring the timing of neural activity in high-order language areas.

Zorik Gekhman, Nadav Oved, Orgad Keller et al. · deepmind

Most works on modeling the conversation history in Conversational Question Answering (CQA) report a single main result on a common CQA benchmark. While existing models show impressive results on CQA leaderboards, it remains unclear whether they are robust to shifts in setting (sometimes to more realistic ones), training data size (e.g. from large to small sets) and domain. In this work, we design and conduct the first large-scale robustness study of history modeling approaches for CQA. We find that high benchmark scores do not necessarily translate to strong robustness, and that various methods can perform extremely differently under different settings. Equipped with the insights from our study, we design a novel prompt-based history modeling approach, and demonstrate its strong robustness across various settings. Our approach is inspired by existing methods that highlight historic answers in the passage. However, instead of highlighting by modifying the passage token embeddings, we add textual prompts directly in the passage text. Our approach is simple, easy-to-plug into practically any model, and highly effective, thus we recommend it as a starting point for future model developers. We also hope that our study and insights will raise awareness to the importance of robustness-focused evaluation, in addition to obtaining high leaderboard scores, leading to better CQA systems.

Tomer Volk, Eyal Ben-David, Ohad Amosy et al. · nvidia

As Natural Language Processing (NLP) algorithms continually achieve new milestones, out-of-distribution generalization remains a significant challenge. This paper addresses the issue of multi-source adaptation for unfamiliar domains: We leverage labeled data from multiple source domains to generalize to unknown target domains at training. Our innovative framework employs example-based Hypernetwork adaptation: a T5 encoder-decoder initially generates a unique signature from an input example, embedding it within the source domains' semantic space. This signature is subsequently utilized by a Hypernetwork to generate the task classifier's weights. We evaluated our method across two tasks - sentiment classification and natural language inference - in 29 adaptation scenarios, where it outpaced established algorithms. In an advanced version, the signature also enriches the input example's representation. We also compare our finetuned architecture to few-shot GPT-3, demonstrating its effectiveness in essential use cases. To our knowledge, this marks the first application of Hypernetworks to the adaptation for unknown domains.

Nirit Nussbaum-Hoffer, Nitay Calderon, Liat Ein-Dor et al.

Causal graphs provide a high-level language for making mechanisms transparent. Recent work uses Large Language Models (LLMs) to recover causal graphs of external-world processes. Instead, in this paper, we use causal graphs to model LLM inference itself, providing stakeholders with a transparent view of how the model perceives and organizes high-level concepts to produce a prediction. We propose a four-phase method for constructing such graphs. Given a target LLM and a set of textual examples, our method discovers class-discriminative, human-interpretable concepts and maps each input to LLM-perceived concept states. We then introduce an MCMC-inspired counterfactual augmentation procedure that expands the sparse observational data through chains of counterfactuals. This enables stable causal discovery with $σ$-CG, yielding informative, interpretable graphs. We apply our method to three LLMs across disease diagnosis, sentiment analysis, and LLM-as-a-judge classification tasks. We evaluate the learned graphs for predictive fidelity and structural stability, and the MCMC-inspired augmentation for convergence and downstream utility. Our results show that the discovered causal graphs capture meaningful dependencies consistent with LLMs' reasoning. Together, this paper provides a foundation for concept-level explainability of LLMs.

Eldar David Abraham, Karel D'Oosterlinck, Amir Feder et al.

The increasing size and complexity of modern ML systems has improved their predictive capabilities but made their behavior harder to explain. Many techniques for model explanation have been developed in response, but we lack clear criteria for assessing these techniques. In this paper, we cast model explanation as the causal inference problem of estimating causal effects of real-world concepts on the output behavior of ML models given actual input data. We introduce CEBaB, a new benchmark dataset for assessing concept-based explanation methods in Natural Language Processing (NLP). CEBaB consists of short restaurant reviews with human-generated counterfactual reviews in which an aspect (food, noise, ambiance, service) of the dining experience was modified. Original and counterfactual reviews are annotated with multiply-validated sentiment ratings at the aspect-level and review-level. The rich structure of CEBaB allows us to go beyond input features to study the effects of abstract, real-world concepts on model behavior. We use CEBaB to compare the quality of a range of concept-based explanation methods covering different assumptions and conceptions of the problem, and we seek to establish natural metrics for comparative assessments of these methods.

Guy Rotman, Roi Reichart

Multi-task learning, in which several tasks are jointly learned by a single model, allows NLP models to share information from multiple annotations and may facilitate better predictions when the tasks are inter-related. This technique, however, requires annotating the same text with multiple annotation schemes which may be costly and laborious. Active learning (AL) has been demonstrated to optimize annotation processes by iteratively selecting unlabeled examples whose annotation is most valuable for the NLP model. Yet, multi-task active learning (MT-AL) has not been applied to state-of-the-art pre-trained Transformer-based NLP models. This paper aims to close this gap. We explore various multi-task selection criteria in three realistic multi-task scenarios, reflecting different relations between the participating tasks, and demonstrate the effectiveness of multi-task compared to single-task selection. Our results suggest that MT-AL can be effectively used in order to minimize annotation efforts for multi-task NLP models.

Mor Ventura, Eyal Ben-David, Anna Korhonen et al.

Text-To-Image (TTI) models, such as DALL-E and StableDiffusion, have demonstrated remarkable prompt-based image generation capabilities. Multilingual encoders may have a substantial impact on the cultural agency of these models, as language is a conduit of culture. In this study, we explore the cultural perception embedded in TTI models by characterizing culture across three hierarchical tiers: cultural dimensions, cultural domains, and cultural concepts. Based on this ontology, we derive prompt templates to unlock the cultural knowledge in TTI models, and propose a comprehensive suite of evaluation techniques, including intrinsic evaluations using the CLIP space, extrinsic evaluations with a Visual-Question-Answer (VQA) model and human assessments, to evaluate the cultural content of TTI-generated images. To bolster our research, we introduce the CulText2I dataset, derived from six diverse TTI models and spanning ten languages. Our experiments provide insights regarding Do, What, Which and How research questions about the nature of cultural encoding in TTI models, paving the way for cross-cultural applications of these models.

Amir Feder, Guy Horowitz, Yoav Wald et al.

Accurately predicting the relevance of items to users is crucial to the success of many social platforms. Conventional approaches train models on logged historical data; but recommendation systems, media services, and online marketplaces all exhibit a constant influx of new content -- making relevancy a moving target, to which standard predictive models are not robust. In this paper, we propose a learning framework for relevance prediction that is robust to changes in the data distribution. Our key observation is that robustness can be obtained by accounting for how users causally perceive the environment. We model users as boundedly-rational decision makers whose causal beliefs are encoded by a causal graph, and show how minimal information regarding the graph can be used to contend with distributional changes. Experiments in multiple settings demonstrate the effectiveness of our approach.

Itai Gat, Nitay Calderon, Roi Reichart et al.

Contemporary predictive models are hard to interpret as their deep nets exploit numerous complex relations between input elements. This work suggests a theoretical framework for model interpretability by measuring the contribution of relevant features to the functional entropy of the network with respect to the input. We rely on the log-Sobolev inequality that bounds the functional entropy by the functional Fisher information with respect to the covariance of the data. This provides a principled way to measure the amount of information contribution of a subset of features to the decision function. Through extensive experiments, we show that our method surpasses existing interpretability sampling-based methods on various data signals such as image, text, and audio.

Yair Gat, Nitay Calderon, Amir Feder et al.

Causal explanations of the predictions of NLP systems are essential to ensure safety and establish trust. Yet, existing methods often fall short of explaining model predictions effectively or efficiently and are often model-specific. In this paper, we address model-agnostic explanations, proposing two approaches for counterfactual (CF) approximation. The first approach is CF generation, where a large language model (LLM) is prompted to change a specific text concept while keeping confounding concepts unchanged. While this approach is demonstrated to be very effective, applying LLM at inference-time is costly. We hence present a second approach based on matching, and propose a method that is guided by an LLM at training-time and learns a dedicated embedding space. This space is faithful to a given causal graph and effectively serves to identify matches that approximate CFs. After showing theoretically that approximating CFs is required in order to construct faithful explanations, we benchmark our approaches and explain several models, including LLMs with billions of parameters. Our empirical results demonstrate the excellent performance of CF generation models as model-agnostic explainers. Moreover, our matching approach, which requires far less test-time resources, also provides effective explanations, surpassing many baselines. We also find that Top-K techniques universally improve every tested method. Finally, we showcase the potential of LLMs in constructing new benchmarks for model explanation and subsequently validate our conclusions. Our work illuminates new pathways for efficient and accurate approaches to interpreting NLP systems.

Alon Berliner, Guy Rotman, Yossi Adi et al.

Discrete variational auto-encoders (VAEs) are able to represent semantic latent spaces in generative learning. In many real-life settings, the discrete latent space consists of high-dimensional structures, and propagating gradients through the relevant structures often requires enumerating over an exponentially large latent space. Recently, various approaches were devised to propagate approximated gradients without enumerating over the space of possible structures. In this work, we use Natural Evolution Strategies (NES), a class of gradient-free black-box optimization algorithms, to learn discrete structured VAEs. The NES algorithms are computationally appealing as they estimate gradients with forward pass evaluations only, thus they do not require to propagate gradients through their discrete structures. We demonstrate empirically that optimizing discrete structured VAEs using NES is as effective as gradient-based approximations. Lastly, we prove NES converges for non-Lipschitz functions as appear in discrete structured VAEs.

Yael Badian, Yaakov Ophir, Refael Tikochinski et al.

The promising research on Artificial Intelligence usages in suicide prevention has principal gaps, including black box methodologies, inadequate outcome measures, and scarce research on non-verbal inputs, such as social media images (despite their popularity today, in our digital era). This study addresses these gaps and combines theory-driven and bottom-up strategies to construct a hybrid and interpretable prediction model of valid suicide risk from images. The lead hypothesis was that images contain valuable information about emotions and interpersonal relationships, two central concepts in suicide-related treatments and theories. The dataset included 177,220 images by 841 Facebook users who completed a gold-standard suicide scale. The images were represented with CLIP, a state-of-the-art algorithm, which was utilized, unconventionally, to extract predefined features that served as inputs to a simple logistic-regression prediction model (in contrast to complex neural networks). The features addressed basic and theory-driven visual elements using everyday language (e.g., bright photo, photo of sad people). The results of the hybrid model (that integrated theory-driven and bottom-up methods) indicated high prediction performance that surpassed common bottom-up algorithms, thus providing a first proof that images (alone) can be leveraged to predict validated suicide risk. Corresponding with the lead hypothesis, at-risk users had images with increased negative emotions and decreased belonginess. The results are discussed in the context of non-verbal warning signs of suicide. Notably, the study illustrates the advantages of hybrid models in such complicated tasks and provides simple and flexible prediction strategies that could be utilized to develop real-life monitoring tools of suicide.

Eyal Ben-David, Yftah Ziser, Roi Reichart

Natural language processing (NLP) algorithms are rapidly improving but often struggle when applied to out-of-distribution examples. A prominent approach to mitigate the domain gap is domain adaptation, where a model trained on a source domain is adapted to a new target domain. We present a new learning setup, ``domain adaptation from scratch'', which we believe to be crucial for extending the reach of NLP to sensitive domains in a privacy-preserving manner. In this setup, we aim to efficiently annotate data from a set of source domains such that the trained model performs well on a sensitive target domain from which data is unavailable for annotation. Our study compares several approaches for this challenging setup, ranging from data selection and domain adaptation algorithms to active learning paradigms, on two NLP tasks: sentiment analysis and Named Entity Recognition. Our results suggest that using the abovementioned approaches eases the domain gap, and combining them further improves the results.

Nitay Calderon, Roi Reichart

Recent advancements in NLP systems, particularly with the introduction of LLMs, have led to widespread adoption of these systems by a broad spectrum of users across various domains, impacting decision-making, the job market, society, and scientific research. This surge in usage has led to an explosion in NLP model interpretability and analysis research, accompanied by numerous technical surveys. Yet, these surveys often overlook the needs and perspectives of explanation stakeholders. In this paper, we address three fundamental questions: Why do we need interpretability, what are we interpreting, and how? By exploring these questions, we examine existing interpretability paradigms, their properties, and their relevance to different stakeholders. We further explore the practical implications of these paradigms by analyzing trends from the past decade across multiple research fields. To this end, we retrieved thousands of papers and employed an LLM to characterize them. Our analysis reveals significant disparities between NLP developers and non-developer users, as well as between research fields, underscoring the diverse needs of stakeholders. For example, explanations of internal model components are rarely used outside the NLP field. We hope this paper informs the future design, development, and application of methods that align with the objectives and requirements of various stakeholders.

Ohad Amosy, Tomer Volk, Eilam Shapira et al.

We address the challenge of building task-agnostic classifiers using only text descriptions, demonstrating a unified approach to image classification, 3D point cloud classification, and action recognition from scenes. Unlike approaches that learn a fixed representation of the output classes, we generate at inference time a model tailored to a query classification task. To generate task-based zero-shot classifiers, we train a hypernetwork that receives class descriptions and outputs a multi-class model. The hypernetwork is designed to be equivariant with respect to the set of descriptions and the classification layer, thus obeying the symmetries of the problem and improving generalization. Our approach generates non-linear classifiers, handles rich textual descriptions, and may be adapted to produce lightweight models efficient enough for on-device applications. We evaluate this approach in a series of zero-shot classification tasks, for image, point-cloud, and action recognition, using a range of text descriptions: From single words to rich descriptions. Our results demonstrate strong improvements over previous approaches, showing that zero-shot learning can be applied with little training data. Furthermore, we conduct an analysis with foundational vision and language models, demonstrating that they struggle to generalize when describing what attributes the class lacks.

Nitay Calderon, Roi Reichart, Rotem Dror

The "LLM-as-an-annotator" and "LLM-as-a-judge" paradigms employ Large Language Models (LLMs) as annotators, judges, and evaluators in tasks traditionally performed by humans. LLM annotations are widely used, not only in NLP research but also in fields like medicine, psychology, and social science. Despite their role in shaping study results and insights, there is no standard or rigorous procedure to determine whether LLMs can replace human annotators. In this paper, we propose a novel statistical procedure, the Alternative Annotator Test (alt-test), that requires only a modest subset of annotated examples to justify using LLM annotations. Additionally, we introduce a versatile and interpretable measure for comparing LLM annotators and judges. To demonstrate our procedure, we curated a diverse collection of ten datasets, consisting of language and vision-language tasks, and conducted experiments with six LLMs and four prompting techniques. Our results show that LLMs can sometimes replace humans with closed-source LLMs (such as GPT-4o), outperforming the open-source LLMs we examine, and that prompting techniques yield judges of varying quality. We hope this study encourages more rigorous and reliable practices.

Alon Goldstein, Miriam Havin, Roi Reichart et al.

This paper investigates the problem-solving capabilities of Large Language Models (LLMs) by evaluating their performance on stumpers, unique single-step intuition problems that pose challenges for human solvers but are easily verifiable. We compare the performance of four state-of-the-art LLMs (Davinci-2, Davinci-3, GPT-3.5-Turbo, GPT-4) to human participants. Our findings reveal that the new-generation LLMs excel in solving stumpers and surpass human performance. However, humans exhibit superior skills in verifying solutions to the same problems. This research enhances our understanding of LLMs' cognitive abilities and provides insights for enhancing their problem-solving potential across various domains.

Lotem Peled-Cohen, Roi Reichart

The close link between cognitive decline and language has fostered long-standing collaboration between the NLP and medical communities in dementia research. To examine this, we reviewed over 240 papers applying NLP to dementia-related efforts, drawing from medical, technological, and NLP-focused literature. We identify key research areas, including dementia detection, linguistic biomarker extraction, caregiver support, and patient assistance, showing that half of all papers focus solely on dementia detection using clinical data. Yet, many directions remain unexplored -- artificially degraded language models, synthetic data, digital twins, and more. We highlight gaps and opportunities around trust, scientific rigor, applicability and cross-community collaboration. We raise ethical dilemmas in the field, and highlight the diverse datasets encountered throughout our review -- recorded, written, structured, spontaneous, synthetic, clinical, social media-based, and more. This review aims to inspire more creative, impactful, and rigorous research on NLP for dementia.

Giulia Occhini, Kumiko Tanaka-Ishii, Anna Barford et al.

Artificial intelligence (AI) has the potential to transform healthcare, education, governance and socioeconomic equity, but its benefits remain concentrated in a small number of languages (Bender, 2019; Blasi et al., 2022; Joshi et al., 2020; Ranathunga and de Silva, 2022; Young, 2015). Language AI - the technologies that underpin widely-used conversational systems such as ChatGPT - could provide major benefits if available in people's native languages, yet most of the world's 7,000+ linguistic communities currently lack access and face persistent digital marginalization. Here we present a global longitudinal analysis of social, economic and infrastructural conditions across languages to assess systemic inequalities in language AI. We first analyze the existence of AI resources for 6003 languages. We find that despite efforts of the community to broaden the reach of language technologies (Bapna et al., 2022; Costa-Jussà et al., 2022), the dominance of a handful of languages is exacerbating disparities on an unprecedented scale, with divides widening exponentially rather than narrowing. Further, we contrast the longitudinal diffusion of AI with that of earlier IT technologies, revealing a distinctive hype-driven pattern of spread. To translate our findings into practical insights and guide prioritization efforts, we introduce the Language AI Readiness Index (EQUATE), which maps the state of technological, socio-economic, and infrastructural prerequisites for AI deployment across languages. The index highlights communities where capacity exists but remains underutilized, and provides a framework for accelerating more equitable diffusion of language AI. Our work contributes to setting the baseline for a transition towards more sustainable and equitable language technologies.

Omer Nahum, Asael Sklar, Ariel Goldstein et al.

Motivation is a central driver of human behavior, shaping decisions, goals, and task performance. As large language models (LLMs) become increasingly aligned with human preferences, we ask whether they exhibit something akin to motivation. We examine whether LLMs "report" varying levels of motivation, how these reports relate to their behavior, and whether external factors can influence them. Our experiments reveal consistent and structured patterns that echo human psychology: self-reported motivation aligns with different behavioral signatures, varies across task types, and can be modulated by external manipulations. These findings demonstrate that motivation is a coherent organizing construct for LLM behavior, systematically linking reports, choices, effort, and performance, and revealing motivational dynamics that resemble those documented in human psychology. This perspective deepens our understanding of model behavior and its connection to human-inspired concepts.

Gilat Toker, Nitay Calderon, Ohad Amosy et al.

Concept-based explanations quantify how high-level concepts (e.g., gender or experience) influence model behavior, which is crucial for decision-makers in high-stakes domains. Recent work evaluates the faithfulness of such explanations by comparing them to reference causal effects estimated from counterfactuals. In practice, existing benchmarks rely on costly human-written counterfactuals that serve as an imperfect proxy. To address this, we introduce a framework for constructing datasets containing structural counterfactual pairs: LIBERTy (LLM-based Interventional Benchmark for Explainability with Reference Targets). LIBERTy is grounded in explicitly defined Structured Causal Models (SCMs) of the text generation, interventions on a concept propagate through the SCM until an LLM generates the counterfactual. We introduce three datasets (disease detection, CV screening, and workplace violence prediction) together with a new evaluation metric, order-faithfulness. Using them, we evaluate a wide range of methods across five models and identify substantial headroom for improving concept-based explanations. LIBERTy also enables systematic analysis of model sensitivity to interventions: we find that proprietary LLMs show markedly reduced sensitivity to demographic concepts, likely due to post-training mitigation. Overall, LIBERTy provides a much-needed benchmark for developing faithful explainability methods.

Eliezer Shlomi, Ido Levy, Eilam Shapira et al.

Planning with LLMs is bottlenecked by token-by-token generation and repeated full forward passes, making multi-step lookahead and rollout-based search expensive in latency and compute. We propose EmbedPlan, which replaces autoregressive next-state generation with a lightweight transition model operating in a frozen language embedding space. EmbedPlan encodes natural language state and action descriptions into vectors, predicts the next-state embedding, and retrieves the next state by nearest-neighbor similarity, enabling fast planning computation without fine-tuning the encoder. We evaluate next-state prediction across nine classical planning domains using six evaluation protocols of increasing difficulty: interpolation, plan-variant, extrapolation, multi-domain, cross-domain, and leave-one-out. Results show near-perfect interpolation performance but a sharp degradation when generalization requires transfer to unseen problems or unseen domains; plan-variant evaluation indicates generalization to alternative plans rather than memorizing seen trajectories. Overall, frozen embeddings support within-domain dynamics learning after observing a domain's transitions, while transfer across domain boundaries remains a bottleneck.

Mor Ventura, Roy Hirsch, Yonatan Bitton et al.

Humans naturally communicate through abstract concepts like "mood". However, current image editing benchmarks focus primarily on explicit, literal commands, leaving abstract instructions largely underexplored. In this work, we first formalize the definition and taxonomy of abstract image editing. To measure instruction-following in this challenging domain, we introduce Entity-Rubrics, a framework that breaks down abstract edits into individual, entity-level assessments and achieves strong correlation with human judgment. Alongside this framework, we contribute AbstractEdit, the first benchmark dedicated to abstract image editing across diverse real-world scenes. Evaluating 11 leading models on this dataset reveals a fundamental challenge: standard architectures struggle to balance intent and preservation, commonly defaulting to under-editing or over-editing. Our analysis demonstrates that driving meaningful improvements relies heavily on integrating advanced LLM text encoders and iterative thinking. Looking forward, our entity-based paradigm can generalize beyond assessment to serve as a reward model, enable models to correctly interpret abstract communication, or highlight specific failures in test-time critique loops. Ultimately, we hope this work serves as a stepping stone toward seamless multimodal interaction, closing the gap between rigid machine execution and the natural, open-ended way humans communicate.

Eilam Shapira, Moshe Tennenholtz, Roi Reichart

Post-training alignment optimizes language models to match human preference signals, but this objective is not equivalent to modeling observed human behavior. We compare 120 base-aligned model pairs on more than 10,000 real human decisions in multi-round strategic games - bargaining, persuasion, negotiation, and repeated matrix games. In these settings, base models outperform their aligned counterparts in predicting human choices by nearly 10:1, robustly across model families, prompt formulations, and game configurations. This pattern reverses, however, in settings where human behavior is more likely to follow normative predictions: aligned models dominate on one-shot textbook games across all 12 types tested and on non-strategic lottery choices - and even within the multi-round games themselves, at round one, before interaction history develops. This boundary-condition pattern suggests that alignment induces a normative bias: it improves prediction when human behavior is relatively well captured by normative solutions, but hurts prediction in multi-round strategic settings, where behavior is shaped by descriptive dynamics such as reciprocity, retaliation, and history-dependent adaptation. These results reveal a fundamental trade-off between optimizing models for human use and using them as proxies for human behavior.

Alan Arazi, Eilam Shapira, Shoham Grunblat et al.

Tabular Foundation Models have recently established the state of the art in supervised tabular learning, by leveraging pretraining to learn generalizable representations of numerical and categorical structured data. However, they lack native support for unstructured modalities such as text and image, and rely on frozen, pretrained embeddings to process them. On established Multimodal Tabular Learning benchmarks, we show that tuning the embeddings to the task improves performance. Existing benchmarks, however, often focus on the mere co-occurrence of modalities; this leads to high variance across datasets and masks the benefits of task-specific tuning. To address this gap, we introduce MulTaBench, a benchmark of 40 datasets, split equally between image-tabular and text-tabular tasks. We focus on predictive tasks where the modalities provide complementary predictive signal, and where generic embeddings lose critical information, necessitating Target-Aware Representations that are aligned with the task. Our experimental results demonstrate that the gains from target-aware representation tuning generalize across both text and image modalities, several tabular learners, encoder scales, and embedding dimensions. MulTaBench constitutes the largest image-tabular benchmarking effort to date, spanning high-impact domains such as healthcare and e-commerce. It is designed to enable the research of novel architectures which incorporate joint modeling and target-aware representations, paving the way for the development of novel Multimodal Tabular Foundation Models.

Gioia Blayer, Myung Jun Kim, Félix Lefebvre et al.

Benchmarking tabular learning has revealed the benefit of dedicated architectures, pushing the state of the art. But real-world tables often contain string entries, beyond numbers, and these settings have been understudied due to a lack of a solid benchmarking suite. They lead to new research questions: Are dedicated learners needed, with end-to-end modeling of strings and numbers? Or does it suffice to encode strings as numbers, as with a categorical encoding? And if so, do the resulting tables resemble numerical tabular data, calling for the same learners? To enable these studies, we contribute STRABLE, a benchmarking corpus of 108 tables, all real-world learning problems with strings and numbers across diverse application fields. We run the first large-scale empirical study of tabular learning with strings, evaluating 445 pipelines. These pipelines span end-to-end architectures and modular pipelines, where strings are first encoded, then post-processed, and finally passed to a tabular learner. We find that, because most tables in the wild are categorical-dominant, advanced tabular learners paired with simple string embeddings achieve good predictions at low computational cost. On free-text-dominant tables, large LLM encoders become competitive. Their performance also appears sensitive to post-processing, with differences across LLM families. Finally, we show that STRABLE is a good set of tables to study "string tabular" learning as it leads to generalizable pipeline rankings that are close to the oracle rankings. We thus establish STRABLE as a foundation for research on tabular learning with strings, an important yet understudied area.

Eilam Shapira, Moshe Tennenholtz, Roi Reichart

AI agents negotiate and transact in natural language with unfamiliar counterparts: a buyer bot facing an unknown seller, or a procurement assistant negotiating with a supplier. In such interactions, the counterpart's LLM, prompts, control logic, and rule-based fallbacks are hidden, while each decision can have monetary consequences. We ask whether an agent can predict an unfamiliar counterpart's next decision from a few interactions. To avoid real-world logging confounds, we study this problem in controlled bargaining and negotiation games, formulating it as target-adaptive text-tabular prediction: each decision point is a table row combining structured game state, offer history, and dialogue, while $K$ previous games of the same target agent, i.e., the counterpart being modeled, are provided in the prompt as labeled adaptation examples. Our model is built on a tabular foundation model that represents rows using game-state features and LLM-based text representations, and adds LLM-as-Observer as an additional representation: a small frozen LLM reads the decision-time state and dialogue; its answer is discarded, and its hidden state becomes a decision-oriented feature, making the LLM an encoder rather than a direct few-shot predictor. Training on 13 frontier-LLM agents and testing on 91 held-out scaffolded agents, the full model outperforms direct LLM-as-Predictor prompting and game+text features baselines. Within this tabular model, Observer features contribute beyond the other feature schemes: at $K=16$, they improve response-prediction AUC by about 4 points across both tasks and reduce bargaining offer-prediction error by 14%. These results show that formulating counterpart prediction as a target-adaptive text-tabular task enables effective adaptation, and that hidden LLM representations expose decision-relevant signals that direct prompting does not surface.

Eilam Shapira, Omer Madmon, Reut Apel et al.

Recent advances in Large Language Models (LLMs) have spurred interest in designing LLM-based agents for tasks that involve interaction with human and artificial agents. This paper addresses a key aspect in the design of such agents: predicting human decisions in off-policy evaluation (OPE). We focus on language-based persuasion games, where an expert aims to influence the decision-maker through verbal messages. In our OPE framework, the prediction model is trained on human interaction data collected from encounters with one set of expert agents, and its performance is evaluated on interactions with a different set of experts. Using a dedicated application, we collected a dataset of 87K decisions from humans playing a repeated decision-making game with artificial agents. To enhance off-policy performance, we propose a simulation technique involving interactions across the entire agent space and simulated decision-makers. Our learning strategy yields significant OPE gains, e.g., improving prediction accuracy in the top 15% challenging cases by 7.1%. Our code and the large dataset we collected and generated are submitted as supplementary material and publicly available in our GitHub repository: https://github.com/eilamshapira/HumanChoicePrediction

Dor Ringel, Rotem Dror, Roi Reichart

We present the Structured Weighted Violation MIRA (SWVM), a new structured prediction algorithm that is based on an hybridization between MIRA (Crammer and Singer, 2003) and the structured weighted violations perceptron (SWVP) (Dror and Reichart, 2016). We demonstrate that the concepts developed in (Dror and Reichart, 2016) combined with a powerful structured prediction algorithm can improve performance on sequence labeling tasks. In experiments with syntactic chunking and named entity recognition (NER), the new algorithm substantially outperforms the original MIRA as well as the original structured perceptron and SWVP. Our code is available at https://github.com/dorringel/SWVM.

Edoardo M. Ponti, Ivan Vulić, Ryan Cotterell et al.

Most combinations of NLP tasks and language varieties lack in-domain examples for supervised training because of the paucity of annotated data. How can neural models make sample-efficient generalizations from task-language combinations with available data to low-resource ones? In this work, we propose a Bayesian generative model for the space of neural parameters. We assume that this space can be factorized into latent variables for each language and each task. We infer the posteriors over such latent variables based on data from seen task-language combinations through variational inference. This enables zero-shot classification on unseen combinations at prediction time. For instance, given training data for named entity recognition (NER) in Vietnamese and for part-of-speech (POS) tagging in Wolof, our model can perform accurate predictions for NER in Wolof. In particular, we experiment with a typologically diverse sample of 33 languages from 4 continents and 11 families, and show that our model yields comparable or better results than state-of-the-art, zero-shot cross-lingual transfer methods. Moreover, we demonstrate that approximate Bayesian model averaging results in smoother predictive distributions, whose entropy inversely correlates with accuracy. Hence, the proposed framework also offers robust estimates of prediction uncertainty. Our code is located at github.com/cambridgeltl/parameter-factorization

Amir Taubenfeld, Yaniv Dover, Roi Reichart et al.

The emergence of Large Language Models (LLMs), has opened exciting possibilities for constructing computational simulations designed to replicate human behavior accurately. Current research suggests that LLM-based agents become increasingly human-like in their performance, sparking interest in using these AI agents as substitutes for human participants in behavioral studies. However, LLMs are complex statistical learners without straightforward deductive rules, making them prone to unexpected behaviors. Hence, it is crucial to study and pinpoint the key behavioral distinctions between humans and LLM-based agents. In this study, we highlight the limitations of LLMs in simulating human interactions, particularly focusing on LLMs' ability to simulate political debates on topics that are important aspects of people's day-to-day lives and decision-making processes. Our findings indicate a tendency for LLM agents to conform to the model's inherent social biases despite being directed to debate from certain political perspectives. This tendency results in behavioral patterns that seem to deviate from well-established social dynamics among humans. We reinforce these observations using an automatic self-fine-tuning method, which enables us to manipulate the biases within the LLM and demonstrate that agents subsequently align with the altered biases. These results underscore the need for further research to develop methods that help agents overcome these biases, a critical step toward creating more realistic simulations.

Eilam Shapira, Roi Reichart, Moshe Tennenholtz

The integration of AI agents into economic markets fundamentally alters the landscape of strategic interaction. We investigate the economic implications of expanding the set of available technologies in three canonical game-theoretic settings: bargaining (resource division), negotiation (asymmetric information trade), and persuasion (strategic information transmission). We find that simply increasing the choice of AI delegates can drastically shift equilibrium payoffs and regulatory outcomes, often creating incentives for regulators to proactively develop and release technologies. Conversely, we identify a strategic phenomenon termed the "Poisoned Apple" effect: an agent may release a new technology, which neither they nor their opponent ultimately uses, solely to manipulate the regulator's choice of market design in their favor. This strategic release improves the releaser's welfare at the expense of their opponent and the regulator's fairness objectives. Our findings demonstrate that static regulatory frameworks are vulnerable to manipulation via technology expansion, necessitating dynamic market designs that adapt to the evolving landscape of AI capabilities.

Shir Lissak, Nitay Calderon, Geva Shenkman et al.

Queer youth face increased mental health risks, such as depression, anxiety, and suicidal ideation. Hindered by negative stigma, they often avoid seeking help and rely on online resources, which may provide incompatible information. Although access to a supportive environment and reliable information is invaluable, many queer youth worldwide have no access to such support. However, this could soon change due to the rapid adoption of Large Language Models (LLMs) such as ChatGPT. This paper aims to comprehensively explore the potential of LLMs to revolutionize emotional support for queers. To this end, we conduct a qualitative and quantitative analysis of LLM's interactions with queer-related content. To evaluate response quality, we develop a novel ten-question scale that is inspired by psychological standards and expert input. We apply this scale to score several LLMs and human comments to posts where queer youth seek advice and share experiences. We find that LLM responses are supportive and inclusive, outscoring humans. However, they tend to be generic, not empathetic enough, and lack personalization, resulting in nonreliable and potentially harmful advice. We discuss these challenges, demonstrate that a dedicated prompt can improve the performance, and propose a blueprint of an LLM-supporter that actively (but sensitively) seeks user context to provide personalized, empathetic, and reliable responses. Our annotated dataset is available for further research.

Lester James V. Miranda, Songbo Hu, Roi Reichart et al.

Where and how language models (LMs) are deployed determines who can benefit from them. However, there are several challenges that prevent effective deployment of LMs in non-English-speaking and hardware constrained communities in the Global South. We call this challenge the last mile: the intersection of multilinguality and edge deployment, where the goals are aligned but the technical requirements often compete. Studying these two fields together is both a need, as linguistically diverse communities often face the most severe infrastructure constraints, and an opportunity, as edge and multilingual NLP research remain largely siloed. To understand the state of the art and the challenges of combining the two areas, we survey 232 papers that tackle this problem across the language modelling pipeline, from data collection to development and deployment. We also discuss open questions and provide actionable recommendations for different stakeholders in the NLP ecosystem. Finally, we hope that this work contributes to the development of inclusive and equitable language technologies.

Omer Nahum, Nitay Calderon, Orgad Keller et al.

NLP benchmarks rely on standardized datasets for training and evaluating models and are crucial for advancing the field. Traditionally, expert annotations ensure high-quality labels; however, the cost of expert annotation does not scale well with the growing demand for larger datasets required by modern models. While crowd-sourcing provides a more scalable solution, it often comes at the expense of annotation precision and consistency. Recent advancements in large language models (LLMs) offer new opportunities to enhance the annotation process, particularly for detecting label errors in existing datasets. In this work, we consider the recent approach of LLM-as-a-judge, leveraging an ensemble of LLMs to flag potentially mislabeled examples. We conduct a case study on four factual consistency datasets from the TRUE benchmark, spanning diverse NLP tasks, and on SummEval, which uses Likert-scale ratings of summary quality across multiple dimensions. We empirically analyze the labeling quality of existing datasets and compare expert, crowd-sourced, and LLM-based annotations in terms of the agreement, label quality, and efficiency, demonstrating the strengths and limitations of each annotation method. Our findings reveal a substantial number of label errors, which, when corrected, induce a significant upward shift in reported model performance. This suggests that many of the LLMs' so-called mistakes are due to label errors rather than genuine model failures. Additionally, we discuss the implications of mislabeled data and propose methods to mitigate them in training to improve performance.

Eilam Shapira, Omer Madmon, Roi Reichart et al.

Human choice prediction in economic contexts is crucial for applications in marketing, finance, public policy, and more. This task, however, is often constrained by the difficulties in acquiring human choice data. With most experimental economics studies focusing on simple choice settings, the AI community has explored whether LLMs can substitute for humans in these predictions and examined more complex experimental economics settings. However, a key question remains: can LLMs generate training data for human choice prediction? We explore this in language-based persuasion games, a complex economic setting involving natural language in strategic interactions. Our experiments show that models trained on LLM-generated data can effectively predict human behavior in these games and even outperform models trained on actual human data. Beyond data generation, we investigate the dual role of LLMs as both data generators and predictors, introducing a comprehensive empirical study on the effectiveness of utilizing LLMs for data generation, human choice prediction, or both. We then utilize our choice prediction framework to analyze how strategic factors shape decision-making, showing that interaction history (rather than linguistic sentiment alone) plays a key role in predicting human decision-making in repeated interactions. Particularly, when LLMs capture history-dependent decision patterns similarly to humans, their predictive success improves substantially. Finally, we demonstrate the robustness of our findings across alternative persuasion-game settings, highlighting the broader potential of using LLM-generated data to model human decision-making.

Itay Nakash, Nitay Calderon, Eyal Ben David et al.

Large Language Models (LLMs) have shown impressive versatility as general purpose models. However, their broad applicability comes at a high-cost computational overhead, particularly in auto-regressive decoding where each step requires a forward pass. In domain-specific settings, general-purpose capabilities are unnecessary and can be exchanged for efficiency. In this work, we take a novel perspective on domain adaptation, reducing latency and computational costs by adapting the vocabulary to focused domains of interest. We introduce AdaptiVocab, an end-to-end approach for vocabulary adaptation, designed to enhance LLM efficiency in low-resource domains. AdaptiVocab can be applied to any tokenizer and architecture, modifying the vocabulary by replacing tokens with domain-specific n-gram-based tokens, thereby reducing the number of tokens required for both input processing and output generation. AdaptiVocab initializes new n-token embeddings using an exponentially weighted combination of existing embeddings and employs a lightweight fine-tuning phase that can be efficiently performed on a single GPU. We evaluate two 7B LLMs across three niche domains, assessing efficiency, generation quality, and end-task performance. Our results show that AdaptiVocab reduces token usage by over 25% without compromising performance

Shir Lissak, Yaakov Ophir, Refael Tikochinski et al.

Background: Recent advancements in Artificial Intelligence (AI) contributed significantly to suicide assessment, however, our theoretical understanding of this complex behavior is still limited. Objective: This study aimed to harness AI methodologies to uncover hidden risk factors that trigger or aggravate suicide behaviors. Method: The primary dataset included 228,052 Facebook postings by 1,006 users who completed the gold-standard Columbia Suicide Severity Rating Scale. This dataset was analyzed using a bottom-up research pipeline without a-priory hypotheses and its findings were validated using a top-down analysis of a new dataset. This secondary dataset included responses by 1,062 participants to the same suicide scale as well as to well-validated scales measuring depression and boredom. Results: An almost fully automated, AI-guided research pipeline resulted in four Facebook topics that predicted the risk of suicide, of which the strongest predictor was boredom. A comprehensive literature review using APA PsycInfo revealed that boredom is rarely perceived as a unique risk factor of suicide. A complementing top-down path analysis of the secondary dataset uncovered an indirect relationship between boredom and suicide, which was mediated by depression. An equivalent mediated relationship was observed in the primary Facebook dataset as well. However, here, a direct relationship between boredom and suicide risk was also observed. Conclusions: Integrating AI methods allowed the discovery of an under-researched risk factor of suicide. The study signals boredom as a maladaptive 'ingredient' that might trigger suicide behaviors, regardless of depression. Further studies are recommended to direct clinicians' attention to this burdening, and sometimes existential experience.

Liam Hazan, Gili Focht, Naama Gavrielov et al.

Automatic conversion of free-text radiology reports into structured data using Natural Language Processing (NLP) techniques is crucial for analyzing diseases on a large scale. While effective for tasks in widely spoken languages like English, generative large language models (LLMs) typically underperform with less common languages and can pose potential risks to patient privacy. Fine-tuning local NLP models is hindered by the skewed nature of real-world medical datasets, where rare findings represent a significant data imbalance. We introduce SMP-BERT, a novel prompt learning method that leverages the structured nature of reports to overcome these challenges. In our studies involving a substantial collection of Crohn's disease radiology reports in Hebrew (over 8,000 patients and 10,000 reports), SMP-BERT greatly surpassed traditional fine-tuning methods in performance, notably in detecting infrequent conditions (AUC: 0.99 vs 0.94, F1: 0.84 vs 0.34). SMP-BERT empowers more accurate AI diagnostics available for low-resource languages.

Yehonatan Peisakhovsky, Zorik Gekhman, Yosi Mass et al. · deepmind

Context-grounded hallucinations are cases where model outputs contain information not verifiable against the source text. We study the applicability of LLMs for localizing such hallucinations, as a more practical alternative to existing complex evaluation pipelines. In the absence of established benchmarks for meta-evaluation of hallucinations localization, we construct one tailored to LLMs, involving a challenging human annotation of over 1,000 examples. We complement the benchmark with an LLM-based evaluation protocol, verifying its quality in a human evaluation. Since existing representations of hallucinations limit the types of errors that can be expressed, we propose a new representation based on free-form textual descriptions, capturing the full range of possible errors. We conduct a comprehensive study, evaluating four large-scale LLMs, which highlights the benchmark's difficulty, as the best model achieves an F1 score of only 0.67. Through careful analysis, we offer insights into optimal prompting strategies for the task and identify the main factors that make it challenging for LLMs: (1) a tendency to incorrectly flag missing details as inconsistent, despite being instructed to check only facts in the output; and (2) difficulty with outputs containing factually correct information absent from the source - and thus not verifiable - due to alignment with the model's parametric knowledge.

Lotem Peled-Cohen, Maya Zadok, Nitay Calderon et al.

Cognitive decline often surfaces in language years before diagnosis. It is frequently non-experts, such as those closest to the patient, who first sense a change and raise concern. As LLMs become integrated into daily communication and used over prolonged periods, it may even be an LLM that notices something is off. But what exactly do they notice--and should be noticing--when making that judgment? This paper investigates how dementia is perceived through language by non-experts. We presented transcribed picture descriptions to non-expert humans and LLMs, asking them to intuitively judge whether each text was produced by someone healthy or with dementia. We introduce an explainable method that uses LLMs to extract high-level, expert-guided features representing these picture descriptions, and use logistic regression to model human and LLM perceptions and compare with clinical diagnoses. Our analysis reveals that human perception of dementia is inconsistent and relies on a narrow, and sometimes misleading, set of cues. LLMs, by contrast, draw on a richer, more nuanced feature set that aligns more closely with clinical patterns. Still, both groups show a tendency toward false negatives, frequently overlooking dementia cases. Through our interpretable framework and the insights it provides, we hope to help non-experts better recognize the linguistic signs that matter.

Alan Arazi, Eilam Shapira, Roi Reichart

While deep learning has achieved remarkable success across many domains, it has historically underperformed on tabular learning tasks, which remain dominated by gradient boosting decision trees. However, recent advancements are paving the way for Tabular Foundation Models, which can leverage real-world knowledge and generalize across diverse datasets, particularly when the data contains free-text. Although incorporating language model capabilities into tabular tasks has been explored, most existing methods utilize static, target-agnostic textual representations, limiting their effectiveness. We introduce TabSTAR: a Tabular Foundation Model with Semantically Target-Aware Representations. TabSTAR is designed to enable transfer learning on tabular data with textual features, with an architecture free of dataset-specific parameters. It unfreezes a pretrained text encoder and takes as input target tokens, which provide the model with the context needed to learn task-specific embeddings. TabSTAR achieves state-of-the-art performance for both medium- and large-sized datasets across known benchmarks of classification tasks with text features, and its pretraining phase exhibits scaling laws in the number of datasets, offering a pathway for further performance improvements.

Kajetan Dymkiewicz, Ivan Vulic, Helen Yannakoudakis et al.

Large language models (LLMs) perform strongly across tasks and languages, yet how improvements in one task or language affect other tasks and languages and their combinations remains poorly understood. We conduct a controlled PEFT/LoRA study across multiple open-weight LLM families and sizes, treating task and language as transfer axes while conditioning on model family and size; we fine-tune each model on a single task-language source and measure transfer as the percentage-point change versus its baseline score when evaluated on all other task-language target pairs. We decompose transfer into (i) Matched-Task (Cross-Language), (ii) Matched-Language (Cross-Task), and (iii) Cross-Task (Cross-Language) regimes. We uncover two consistent general patterns. First, a pronounced on-task vs. off-task asymmetry: Matched-Task (Cross-Language) transfer is reliably positive, whereas off-task transfer often incurs collateral degradation. Second, a stable donor-recipient structure across languages and tasks (hub donors vs. brittle recipients). We outline implications for risk-aware fine-tuning and model specialisation.

Adi Simhi, Jonathan Herzig, Itay Itzhak et al. · deepmind

Hallucinations in LLMs present a critical barrier to their reliable usage. Existing research usually categorizes hallucination by their external properties rather than by the LLMs' underlying internal properties. This external focus overlooks that hallucinations may require tailored mitigation strategies based on their underlying mechanism. We propose a framework for categorizing hallucinations along two axes: knowledge and certainty. Since parametric knowledge and certainty may vary across models, our categorization method involves a model-specific dataset construction process that differentiates between those types of hallucinations. Along the knowledge axis, we distinguish between hallucinations caused by a lack of knowledge and those occurring despite the model having the knowledge of the correct response. To validate our framework along the knowledge axis, we apply steering mitigation, which relies on the existence of parametric knowledge to manipulate model activations. This addresses the lack of existing methods to validate knowledge categorization by showing a significant difference between the two hallucination types. We further analyze the distinct knowledge and hallucination patterns between models, showing that different hallucinations do occur despite shared parametric knowledge. Turning to the certainty axis, we identify a particularly concerning subset of hallucinations where models hallucinate with certainty despite having the correct knowledge internally. We introduce a new evaluation metric to measure the effectiveness of mitigation methods on this subset, revealing that while some methods perform well on average, they fail disproportionately on these critical cases. Our findings highlight the importance of considering both knowledge and certainty in hallucination analysis and call for targeted mitigation approaches that consider the hallucination underlying factors.

Mor Ventura, Michael Toker, Or Patashnik et al.

Text-to-Image (T2I) models have advanced rapidly, yet they remain vulnerable to semantic leakage, the unintended transfer of semantically related features between distinct entities. Existing mitigation strategies are often optimization-based or dependent on external inputs. We introduce DeLeaker, a lightweight, optimization-free inference-time approach that mitigates leakage by directly intervening on the model's attention maps. Throughout the diffusion process, DeLeaker dynamically reweights attention maps to suppress excessive cross-entity interactions while strengthening the identity of each entity. To support systematic evaluation, we introduce SLIM (Semantic Leakage in IMages), the first dataset dedicated to semantic leakage, comprising 1,130 human-verified samples spanning diverse scenarios, together with a novel automatic evaluation framework. Experiments demonstrate that DeLeaker consistently outperforms all baselines, even when they are provided with external information, achieving effective leakage mitigation without compromising fidelity or quality. These results underscore the value of attention control and pave the way for more semantically precise T2I models.

Ofir Azachi, Kfir Eliyahu, Eyal El Ani et al.

Hallucinations of vision-language models (VLMs), which are misalignments between visual content and generated text, undermine the reliability of VLMs. One common approach for detecting them employs the same VLM, or a different one, to assess generated outputs. This process is computationally intensive and increases model latency. In this paper, we explore an efficient on-the-fly method for hallucination detection by training traditional ML models over signals based on the VLM's next-token probabilities (NTPs). NTPs provide a direct quantification of model uncertainty. We hypothesize that high uncertainty (i.e., a low NTP value) is strongly associated with hallucinations. To test this, we introduce a dataset of 1,400 human-annotated statements derived from VLM-generated content, each labeled as hallucinated or not, and use it to test our NTP-based lightweight method. Our results demonstrate that NTP-based features are valuable predictors of hallucinations, enabling fast and simple ML models to achieve performance comparable to that of strong VLMs. Furthermore, augmenting these NTPs with linguistic NTPs, computed by feeding only the generated text back into the VLM, enhances hallucination detection performance. Finally, integrating hallucination prediction scores from VLMs into the NTP-based models led to better performance than using either VLMs or NTPs alone. We hope this study paves the way for simple, lightweight solutions that enhance the reliability of VLMs.

Ilanit Sobol, Shir Lissak, Refael Tikochinski et al.

Suicide remains a leading cause of death in Western countries, underscoring the need for new research approaches. As social media becomes central to daily life, digital footprints offer valuable insight into suicidal behavior. Focusing on individuals who attempted suicide while uploading videos to their channels, we investigate: How do suicidal behaviors manifest on YouTube, and how do they differ from expert knowledge? We applied complementary approaches: computational bottom-up, hybrid, and expert-driven top-down, on a novel longitudinal dataset of 181 YouTube channels from individuals with life-threatening attempts, alongside 134 control channels. In the bottom-up approach, we applied LLM-based topic modeling to identify behavioral indicators. Of 166 topics, five were associated with suicide-attempt, with two also showing temporal attempt-related changes ($p<.01$) - Mental Health Struggles ($+0.08$)* and YouTube Engagement ($+0.1$)*. In the hybrid approach, a clinical expert reviewed LLM-derived topics and flagged 19 as suicide-related. However, none showed significant attempt-related temporal effects beyond those identified bottom-up. Notably, YouTube Engagement, a platform-specific indicator, was not flagged by the expert, underscoring the value of bottom-up discovery. In the top-down approach, psychological assessment of suicide attempt narratives revealed that the only significant difference between individuals who attempted before and those attempted during their upload period was the motivation to share this experience: the former aimed to Help Others ($β=-1.69$, $p<.01$), while the latter framed it as part of their Personal Recovery ($β=1.08$, $p<.01$). By integrating these approaches, we offer a nuanced understanding of suicidality, bridging digital behavior and clinical insights. * Within-group changes in relation to the suicide attempt.

Zorik Gekhman, Eyal Ben David, Hadas Orgad et al.

This work presents a framework for assessing whether large language models (LLMs) encode more factual knowledge in their parameters than what they express in their outputs. While a few studies hint at this possibility, none has clearly defined or demonstrated this phenomenon. We first propose a formal definition of knowledge, quantifying it for a given question as the fraction of correct-incorrect answer pairs where the correct one is ranked higher. This gives rise to external and internal knowledge, depending on the information used to score individual answer candidates: either the model's observable token-level probabilities or its intermediate computations. Hidden knowledge arises when internal knowledge exceeds external knowledge. We then present a case study, applying this framework to three popular open-weights LLMs in a closed-book QA setup. Our results indicate that: (1) LLMs consistently encode more factual knowledge internally than what they express externally, with an average relative gap of 40%. (2) Surprisingly, some knowledge is so deeply hidden that a model can internally know an answer perfectly, yet fail to generate it even once, despite large-scale repeated sampling of 1,000 answers. This reveals fundamental limitations in the generation capabilities of LLMs, which (3) put a practical constraint on scaling test-time compute via repeated answer sampling in closed-book QA: significant performance improvements remain inaccessible because some answers are practically never sampled, yet if they were, we would be guaranteed to rank them first.

Almog Gueta, Amir Feder, Zorik Gekhman et al.

This study empirically tests the $\textit{Narrative Economics}$ hypothesis, which posits that narratives (ideas that are spread virally and affect public beliefs) can influence economic fluctuations. We introduce two curated datasets containing posts from X (formerly Twitter) which capture economy-related narratives (Data will be shared upon paper acceptance). Employing Natural Language Processing (NLP) methods, we extract and summarize narratives from the tweets. We test their predictive power for $\textit{macroeconomic}$ forecasting by incorporating the tweets' or the extracted narratives' representations in downstream financial prediction tasks. Our work highlights the challenges in improving macroeconomic models with narrative data, paving the way for the research community to realistically address this important challenge. From a scientific perspective, our investigation offers valuable insights and NLP tools for narrative extraction and summarization using Large Language Models (LLMs), contributing to future research on the role of narratives in economics.