Daily Papers

Recent work has shown that models trained to the same objective, and which achieve similar measures of accuracy on consistent test data, may nonetheless behave very differently on individual predictions. This inconsistency is undesirable in high-stakes contexts, such as medical diagnosis and finance. We show that this inconsistent behavior extends beyond predictions to feature attributions, which may likewise have negative implications for the intelligibility of a model, and one's ability to find recourse for subjects. We then introduce selective ensembles to mitigate such inconsistencies by applying hypothesis testing to the predictions of a set of models trained using randomly-selected starting conditions; importantly, selective ensembles can abstain in cases where a consistent outcome cannot be achieved up to a specified confidence level. We prove that that prediction disagreement between selective ensembles is bounded, and empirically demonstrate that selective ensembles achieve consistent predictions and feature attributions while maintaining low abstention rates. On several benchmark datasets, selective ensembles reach zero inconsistently predicted points, with abstention rates as low 1.5%.

Large language models (LLMs) chatbots like ChatGPT are increasingly used for mental health support. They offer accessible, therapeutic support but also raise concerns about misinformation, over-reliance, and risks in high-stakes contexts of mental health. We crowdsource large-scale users' posts from six major social media platforms to examine how people discuss their interactions with LLM chatbots across different mental health conditions. Through an LLM-assisted pipeline grounded in Value-Sensitive Design (VSD), we mapped the relationships across user-reported sentiments, mental health conditions, perspectives, and values. Our results reveal that the use of LLM chatbots is condition-specific. Users with neurodivergent conditions (e.g., ADHD, ASD) report strong positive sentiments and instrumental or appraisal support, whereas higher-risk disorders (e.g., schizophrenia, bipolar disorder) show more negative sentiments. We further uncover how user perspectives co-occur with underlying values, such as identity, autonomy, and privacy. Finally, we discuss shifting from "one-size-fits-all" chatbot design toward condition-specific, value-sensitive LLM design.

Digital Forensics and Incident Response (DFIR) involves analyzing digital evidence to support legal investigations. Large Language Models (LLMs) offer new opportunities in DFIR tasks such as log analysis and memory forensics, but their susceptibility to errors and hallucinations raises concerns in high-stakes contexts. Despite growing interest, there is no comprehensive benchmark to evaluate LLMs across both theoretical and practical DFIR domains. To address this gap, we present DFIR-Metric, a benchmark with three components: (1) Knowledge Assessment: a set of 700 expert-reviewed multiple-choice questions sourced from industry-standard certifications and official documentation; (2) Realistic Forensic Challenges: 150 CTF-style tasks testing multi-step reasoning and evidence correlation; and (3) Practical Analysis: 500 disk and memory forensics cases from the NIST Computer Forensics Tool Testing Program (CFTT). We evaluated 14 LLMs using DFIR-Metric, analyzing both their accuracy and consistency across trials. We also introduce a new metric, the Task Understanding Score (TUS), designed to more effectively evaluate models in scenarios where they achieve near-zero accuracy. This benchmark offers a rigorous, reproducible foundation for advancing AI in digital forensics. All scripts, artifacts, and results are available on the project website at https://github.com/DFIR-Metric.

While research on applications and evaluations of explanation methods continues to expand, fairness of the explanation methods concerning disparities in their performance across subgroups remains an often overlooked aspect. In this paper, we address this gap by showing that, across three tasks and five language models, widely used post-hoc feature attribution methods exhibit significant gender disparity with respect to their faithfulness, robustness, and complexity. These disparities persist even when the models are pre-trained or fine-tuned on particularly unbiased datasets, indicating that the disparities we observe are not merely consequences of biased training data. Our results highlight the importance of addressing disparities in explanations when developing and applying explainability methods, as these can lead to biased outcomes against certain subgroups, with particularly critical implications in high-stakes contexts. Furthermore, our findings underscore the importance of incorporating the fairness of explanations, alongside overall model fairness and explainability, as a requirement in regulatory frameworks.

The rapid advancement of artificial intelligence in legal natural language processing demands scalable methods for evaluating text extraction from judicial decisions. This study evaluates 16 unsupervised metrics, including novel formulations, to assess the quality of extracting seven semantic blocks from 1,000 anonymized Russian judicial decisions, validated against 7,168 expert reviews on a 1--5 Likert scale. These metrics, spanning document-based, semantic, structural, pseudo-ground truth, and legal-specific categories, operate without pre-annotated ground truth. Bootstrapped correlations, Lin's concordance correlation coefficient (CCC), and mean absolute error (MAE) reveal that Term Frequency Coherence (Pearson r = 0.540, Lin CCC = 0.512, MAE = 0.127) and Coverage Ratio/Block Completeness (Pearson r = 0.513, Lin CCC = 0.443, MAE = 0.139) best align with expert ratings, while Legal Term Density (Pearson r = -0.479, Lin CCC = -0.079, MAE = 0.394) show strong negative correlations. The LLM Evaluation Score (mean = 0.849, Pearson r = 0.382, Lin CCC = 0.325, MAE = 0.197) showed moderate alignment, but its performance, using gpt-4.1-mini via g4f, suggests limited specialization for legal textse. These findings highlight that unsupervised metrics, including LLM-based approaches, enable scalable screening but, with moderate correlations and low CCC values, cannot fully replace human judgment in high-stakes legal contexts. This work advances legal NLP by providing annotation-free evaluation tools, with implications for judicial analytics and ethical AI deployment.

Frontier model progress is often measured by academic benchmarks, which offer a limited view of performance in real-world professional contexts. Existing evaluations often fail to assess open-ended, economically consequential tasks in high-stakes domains like Legal and Finance, where practical returns are paramount. To address this, we introduce Professional Reasoning Bench (PRBench), a realistic, open-ended, and difficult benchmark of real-world problems in Finance and Law. We open-source its 1,100 expert-authored tasks and 19,356 expert-curated criteria, making it, to our knowledge, the largest public, rubric-based benchmark for both legal and finance domains. We recruit 182 qualified professionals, holding JDs, CFAs, or 6+ years of experience, who contributed tasks inspired by their actual workflows. This process yields significant diversity, with tasks spanning 114 countries and 47 US jurisdictions. Our expert-curated rubrics are validated through a rigorous quality pipeline, including independent expert validation. Subsequent evaluation of 20 leading models reveals substantial room for improvement, with top scores of only 0.39 (Finance) and 0.37 (Legal) on our Hard subsets. We further catalog associated economic impacts of the prompts and analyze performance using human-annotated rubric categories. Our analysis shows that models with similar overall scores can diverge significantly on specific capabilities. Common failure modes include inaccurate judgments, a lack of process transparency and incomplete reasoning, highlighting critical gaps in their reliability for professional adoption.

Governments are increasingly considering integrating autonomous AI agents in high-stakes military and foreign-policy decision-making, especially with the emergence of advanced generative AI models like GPT-4. Our work aims to scrutinize the behavior of multiple AI agents in simulated wargames, specifically focusing on their predilection to take escalatory actions that may exacerbate multilateral conflicts. Drawing on political science and international relations literature about escalation dynamics, we design a novel wargame simulation and scoring framework to assess the escalation risks of actions taken by these agents in different scenarios. Contrary to prior studies, our research provides both qualitative and quantitative insights and focuses on large language models (LLMs). We find that all five studied off-the-shelf LLMs show forms of escalation and difficult-to-predict escalation patterns. We observe that models tend to develop arms-race dynamics, leading to greater conflict, and in rare cases, even to the deployment of nuclear weapons. Qualitatively, we also collect the models' reported reasonings for chosen actions and observe worrying justifications based on deterrence and first-strike tactics. Given the high stakes of military and foreign-policy contexts, we recommend further examination and cautious consideration before deploying autonomous language model agents for strategic military or diplomatic decision-making.

Despite advances in AI's performance and interpretability, AI advisors can undermine experts' decisions and increase the time and effort experts must invest to make decisions. Consequently, AI systems deployed in high-stakes settings often fail to consistently add value across contexts and can even diminish the value that experts alone provide. Beyond harm in specific domains, such outcomes impede progress in research and practice, underscoring the need to understand when and why different AI advisors add or diminish value. To bridge this gap, we stress the importance of assessing the value AI advice brings to real-world contexts when designing and evaluating AI advisors. Building on this perspective, we characterize key pillars -- pathways through which AI advice impacts value -- and develop a framework that incorporates these pillars to create reliable, personalized, and value-adding advisors. Our results highlight the need for system-level, value-driven development of AI advisors that advise selectively, are tailored to experts' unique behaviors, and are optimized for context-specific trade-offs between decision improvements and advising costs. They also reveal how the lack of inclusion of these pillars in the design of AI advising systems may be contributing to the failures observed in practical applications.

Language-driven generative agents have enabled large-scale social simulations with transformative uses, from interpersonal training to aiding global policy-making. However, recent studies indicate that generative agent behaviors often deviate from expert expectations and real-world data--a phenomenon we term the Behavior-Realism Gap. To address this, we introduce a theoretical framework called Persona-Environment Behavioral Alignment (PEBA), formulated as a distribution matching problem grounded in Lewin's behavior equation stating that behavior is a function of the person and their environment. Leveraging PEBA, we propose PersonaEvolve (PEvo), an LLM-based optimization algorithm that iteratively refines agent personas, implicitly aligning their collective behaviors with realistic expert benchmarks within a specified environmental context. We validate PEvo in an active shooter incident simulation we developed, achieving an 84% average reduction in distributional divergence compared to no steering and a 34% improvement over explicit instruction baselines. Results also show PEvo-refined personas generalize to novel, related simulation scenarios. Our method greatly enhances behavioral realism and reliability in high-stakes social simulations. More broadly, the PEBA-PEvo framework provides a principled approach to developing trustworthy LLM-driven social simulations.

Current large language models (LLMs) excel in verifiable domains where outputs can be checked before action but prove less reliable for high-stakes strategic decisions with uncertain outcomes. This gap, driven by mutually reinforcing cognitive biases in both humans and artificial intelligence (AI) systems, threatens the defensibility of valuations and sustainability of investments in the sector. This report describes a framework emerging from systematic qualitative assessment across 7 frontier-grade LLMs and 3 market-facing venture vignettes under time pressure. Detailed prompting specifying decision partnership and explicitly instructing avoidance of sycophancy, confabulation, solution drift, and nihilism achieved initial partnership state but failed to maintain it under operational pressure. Sustaining protective partnership state required an emergent 7-stage calibration sequence, built upon a 4-stage initialization process, within a 5-layer protection architecture enabling bias self-monitoring, human-AI adversarial challenge, partnership state verification, performance degradation detection, and stakeholder protection. Three discoveries resulted: partnership state is achievable through ordered calibration but requires emergent maintenance protocols; reliability degrades when architectural drift and context exhaustion align; and dissolution discipline prevents costly pursuit of fundamentally wrong directions. Cross-model validation revealed systematic performance differences across LLM architectures. This approach demonstrates that human-AI teams can achieve cognitive partnership capable of preventing avoidable regret in high-stakes decisions, addressing return-on-investment expectations that depend on AI systems supporting consequential decision-making without introducing preventable cognitive traps when verification arrives too late.

The proliferation of LLM-based agents has led to increasing deployment of inter-agent collaboration for tasks like scheduling, negotiation, resource allocation etc. In such systems, privacy is critical, as agents often access proprietary tools and domain-specific databases requiring strict confidentiality. This paper examines whether LLM-based agents demonstrate an understanding of contextual privacy. And, if instructed, do these systems preserve inference time user privacy in non-adversarial multi-turn conversation. Existing benchmarks to evaluate contextual privacy in LLM-agents primarily assess single-turn, low-complexity tasks where private information can be easily excluded. We first present a benchmark - MAGPIE comprising 158 real-life high-stakes scenarios across 15 domains. These scenarios are designed such that complete exclusion of private data impedes task completion yet unrestricted information sharing could lead to substantial losses. We then evaluate the current state-of-the-art LLMs on (a) their understanding of contextually private data and (b) their ability to collaborate without violating user privacy. Empirical experiments demonstrate that current models, including GPT-4o and Claude-2.7-Sonnet, lack robust understanding of contextual privacy, misclassifying private data as shareable 25.2\% and 43.6\% of the time. In multi-turn conversations, these models disclose private information in 59.9\% and 50.5\% of cases even under explicit privacy instructions. Furthermore, multi-agent systems fail to complete tasks in 71\% of scenarios. These results underscore that current models are not aligned towards both contextual privacy preservation and collaborative task-solving.

The integration of Large Language Models (LLMs) into healthcare demands a safety paradigm rooted in primum non nocere. However, current alignment techniques rely on generic definitions of harm that fail to capture context-dependent violations, such as administrative fraud and clinical discrimination. To address this, we introduce Medical Malice: a dataset of 214,219 adversarial prompts calibrated to the regulatory and ethical complexities of the Brazilian Unified Health System (SUS). Crucially, the dataset includes the reasoning behind each violation, enabling models to internalize ethical boundaries rather than merely memorizing a fixed set of refusals. Using an unaligned agent (Grok-4) within a persona-driven pipeline, we synthesized high-fidelity threats across seven taxonomies, ranging from procurement manipulation and queue-jumping to obstetric violence. We discuss the ethical design of releasing these "vulnerability signatures" to correct the information asymmetry between malicious actors and AI developers. Ultimately, this work advocates for a shift from universal to context-aware safety, providing the necessary resources to immunize healthcare AI against the nuanced, systemic threats inherent to high-stakes medical environments -- vulnerabilities that represent the paramount risk to patient safety and the successful integration of AI in healthcare systems.

A core challenge for autonomous LLM agents in collaborative settings is balancing robust privacy understanding and preservation alongside task efficacy. Existing privacy benchmarks only focus on simplistic, single-turn interactions where private information can be trivially omitted without affecting task outcomes. In this paper, we introduce MAGPIE (Multi-AGent contextual PrIvacy Evaluation), a novel benchmark of 200 high-stakes tasks designed to evaluate privacy understanding and preservation in multi-agent collaborative, non-adversarial scenarios. MAGPIE integrates private information as essential for task resolution, forcing agents to balance effective collaboration with strategic information control. Our evaluation reveals that state-of-the-art agents, including GPT-5 and Gemini 2.5-Pro, exhibit significant privacy leakage, with Gemini 2.5-Pro leaking up to 50.7% and GPT-5 up to 35.1% of the sensitive information even when explicitly instructed not to. Moreover, these agents struggle to achieve consensus or task completion and often resort to undesirable behaviors such as manipulation and power-seeking (e.g., Gemini 2.5-Pro demonstrating manipulation in 38.2% of the cases). These findings underscore that current LLM agents lack robust privacy understanding and are not yet adequately aligned to simultaneously preserve privacy and maintain effective collaboration in complex environments.

This paper introduces CaseSumm, a novel dataset for long-context summarization in the legal domain that addresses the need for longer and more complex datasets for summarization evaluation. We collect 25.6K U.S. Supreme Court (SCOTUS) opinions and their official summaries, known as "syllabuses." Our dataset is the largest open legal case summarization dataset, and is the first to include summaries of SCOTUS decisions dating back to 1815. We also present a comprehensive evaluation of LLM-generated summaries using both automatic metrics and expert human evaluation, revealing discrepancies between these assessment methods. Our evaluation shows Mistral 7b, a smaller open-source model, outperforms larger models on most automatic metrics and successfully generates syllabus-like summaries. In contrast, human expert annotators indicate that Mistral summaries contain hallucinations. The annotators consistently rank GPT-4 summaries as clearer and exhibiting greater sensitivity and specificity. Further, we find that LLM-based evaluations are not more correlated with human evaluations than traditional automatic metrics. Furthermore, our analysis identifies specific hallucinations in generated summaries, including precedent citation errors and misrepresentations of case facts. These findings demonstrate the limitations of current automatic evaluation methods for legal summarization and highlight the critical role of human evaluation in assessing summary quality, particularly in complex, high-stakes domains. CaseSumm is available at https://huggingface.co/datasets/ChicagoHAI/CaseSumm

Trust is an important factor in people's interactions with AI systems. However, there is a lack of empirical studies examining how real end-users trust or distrust the AI system they interact with. Most research investigates one aspect of trust in lab settings with hypothetical end-users. In this paper, we provide a holistic and nuanced understanding of trust in AI through a qualitative case study of a real-world computer vision application. We report findings from interviews with 20 end-users of a popular, AI-based bird identification app where we inquired about their trust in the app from many angles. We find participants perceived the app as trustworthy and trusted it, but selectively accepted app outputs after engaging in verification behaviors, and decided against app adoption in certain high-stakes scenarios. We also find domain knowledge and context are important factors for trust-related assessment and decision-making. We discuss the implications of our findings and provide recommendations for future research on trust in AI.

Large language models (LLMs) have shown promise in medical question answering, yet they often overlook the domain-specific expertise that professionals depend on, such as the clinical subject areas (e.g., trauma, airway) and the certification level (e.g., EMT, Paramedic). Existing approaches typically apply general-purpose prompting or retrieval strategies without leveraging this structured context, limiting performance in high-stakes settings. We address this gap with EMSQA, an 24.3K-question multiple-choice dataset spanning 10 clinical subject areas and 4 certification levels, accompanied by curated, subject area-aligned knowledge bases (40K documents and 2M tokens). Building on EMSQA, we introduce (i) Expert-CoT, a prompting strategy that conditions chain-of-thought (CoT) reasoning on specific clinical subject area and certification level, and (ii) ExpertRAG, a retrieval-augmented generation pipeline that grounds responses in subject area-aligned documents and real-world patient data. Experiments on 4 LLMs show that Expert-CoT improves up to 2.05% over vanilla CoT prompting. Additionally, combining Expert-CoT with ExpertRAG yields up to a 4.59% accuracy gain over standard RAG baselines. Notably, the 32B expertise-augmented LLMs pass all the computer-adaptive EMS certification simulation exams.

Legal practitioners, particularly those early in their careers, face complex, high-stakes tasks that require adaptive, context-sensitive reasoning. While AI holds promise in supporting legal work, current datasets and models are narrowly focused on isolated subtasks and fail to capture the end-to-end decision-making required in real-world practice. To address this gap, we introduce LawFlow, a dataset of complete end-to-end legal workflows collected from trained law students, grounded in real-world business entity formation scenarios. Unlike prior datasets focused on input-output pairs or linear chains of thought, LawFlow captures dynamic, modular, and iterative reasoning processes that reflect the ambiguity, revision, and client-adaptive strategies of legal practice. Using LawFlow, we compare human and LLM-generated workflows, revealing systematic differences in structure, reasoning flexibility, and plan execution. Human workflows tend to be modular and adaptive, while LLM workflows are more sequential, exhaustive, and less sensitive to downstream implications. Our findings also suggest that legal professionals prefer AI to carry out supportive roles, such as brainstorming, identifying blind spots, and surfacing alternatives, rather than executing complex workflows end-to-end. Building on these findings, we propose a set of design suggestions, rooted in empirical observations, that align AI assistance with human goals of clarity, completeness, creativity, and efficiency, through hybrid planning, adaptive execution, and decision-point support. Our results highlight both the current limitations of LLMs in supporting complex legal workflows and opportunities for developing more collaborative, reasoning-aware legal AI systems. All data and code are available on our project page (https://minnesotanlp.github.io/LawFlow-website/).

Large Language Models (LLMs) face limitations in AI legal and policy applications due to outdated knowledge, hallucinations, and poor reasoning in complex contexts. Retrieval-Augmented Generation (RAG) systems address these issues by incorporating external knowledge, but suffer from retrieval errors, ineffective context integration, and high operational costs. This paper presents the Hybrid Parameter-Adaptive RAG (HyPA-RAG) system, designed for the AI legal domain, with NYC Local Law 144 (LL144) as the test case. HyPA-RAG integrates a query complexity classifier for adaptive parameter tuning, a hybrid retrieval approach combining dense, sparse, and knowledge graph methods, and a comprehensive evaluation framework with tailored question types and metrics. Testing on LL144 demonstrates that HyPA-RAG enhances retrieval accuracy, response fidelity, and contextual precision, offering a robust and adaptable solution for high-stakes legal and policy applications.

The rapid advancement of large language models (LLMs) has significantly impacted various domains, including healthcare and biomedicine. However, the phenomenon of hallucination, where LLMs generate outputs that deviate from factual accuracy or context, poses a critical challenge, especially in high-stakes domains. This paper conducts a scoping study of existing techniques for mitigating hallucinations in knowledge-based task in general and especially for medical domains. Key methods covered in the paper include Retrieval-Augmented Generation (RAG)-based techniques, iterative feedback loops, supervised fine-tuning, and prompt engineering. These techniques, while promising in general contexts, require further adaptation and optimization for the medical domain due to its unique demands for up-to-date, specialized knowledge and strict adherence to medical guidelines. Addressing these challenges is crucial for developing trustworthy AI systems that enhance clinical decision-making and patient safety as well as accuracy of biomedical scientific research.

Recent work has highlighted the culturally-contingent nature of commonsense knowledge. We introduce AMAMMER{epsilon}, a test set of 525 multiple-choice questions designed to evaluate the commonsense knowledge of English LLMs, relative to the cultural contexts of Ghana and the United States. To create AMAMMER{epsilon}, we select a set of multiple-choice questions (MCQs) from existing commonsense datasets and rewrite them in a multi-stage process involving surveys of Ghanaian and U.S. participants. In three rounds of surveys, participants from both pools are solicited to (1) write correct and incorrect answer choices, (2) rate individual answer choices on a 5-point Likert scale, and (3) select the best answer choice from the newly-constructed MCQ items, in a final validation step. By engaging participants at multiple stages, our procedure ensures that participant perspectives are incorporated both in the creation and validation of test items, resulting in high levels of agreement within each pool. We evaluate several off-the-shelf English LLMs on AMAMMER{epsilon}. Uniformly, models prefer answers choices that align with the preferences of U.S. annotators over Ghanaian annotators. Additionally, when test items specify a cultural context (Ghana or the U.S.), models exhibit some ability to adapt, but performance is consistently better in U.S. contexts than Ghanaian. As large resources are devoted to the advancement of English LLMs, our findings underscore the need for culturally adaptable models and evaluations to meet the needs of diverse English-speaking populations around the world.

Science and Engineering fairs offer K-12 students opportunities to engage with authentic STEM practices. Particularly, students are given the chance to experience authentic and open inquiry processes, by defining which themes, questions and approaches will guide their scientific endeavors. In this study, we analyzed data from over 5,000 projects presented at a nationwide science fair in Brazil over the past 20 years using topic modeling to identify the main topics that have driven students' inquiry and design. Our analysis identified a broad range of topics being explored, with significant variations over time, region, and school setting. We argue those results and proposed methodology can not only support further research in the context of science fairs, but also inform instruction and design of contexts-specific resources to support students in open inquiry experiences in different settings.

Recent advancements in large language models (LLM) capable of processing extremely long texts highlight the need for a dedicated evaluation benchmark to assess their long-context capabilities. However, existing methods, like the needle-in-a-haystack test, do not effectively assess whether these models fully utilize contextual information, raising concerns about the reliability of current evaluation techniques. To thoroughly examine the effectiveness of existing benchmarks, we introduce a new metric called information coverage (IC), which quantifies the proportion of the input context necessary for answering queries. Our findings indicate that current benchmarks exhibit low IC; although the input context may be extensive, the actual usable context is often limited. To address this, we present ETHIC, a novel benchmark designed to assess LLMs' ability to leverage the entire context. Our benchmark comprises 2,648 test instances spanning four long-context tasks with high IC scores in the domains of books, debates, medicine, and law. Our evaluations reveal significant performance drops in contemporary LLMs, highlighting a critical challenge in managing long contexts. Our benchmark is available at https://github.com/dmis-lab/ETHIC.

Recent large language models (LLMs) support long contexts ranging from 128K to 1M tokens. A popular method for evaluating these capabilities is the needle-in-a-haystack (NIAH) test, which involves retrieving a "needle" (relevant information) from a "haystack" (long irrelevant context). Extensions of this approach include increasing distractors, fact chaining, and in-context reasoning. However, in these benchmarks, models can exploit existing literal matches between the needle and haystack to simplify the task. To address this, we introduce NoLiMa, a benchmark extending NIAH with a carefully designed needle set, where questions and needles have minimal lexical overlap, requiring models to infer latent associations to locate the needle within the haystack. We evaluate 12 popular LLMs that claim to support contexts of at least 128K tokens. While they perform well in short contexts (<1K), performance degrades significantly as context length increases. At 32K, for instance, 10 models drop below 50% of their strong short-length baselines. Even GPT-4o, one of the top-performing exceptions, experiences a reduction from an almost-perfect baseline of 99.3% to 69.7%. Our analysis suggests these declines stem from the increased difficulty the attention mechanism faces in longer contexts when literal matches are absent, making it harder to retrieve relevant information.

A key component of successfully reading a passage of text is the ability to apply knowledge gained from the passage to a new situation. In order to facilitate progress on this kind of reading, we present ROPES, a challenging benchmark for reading comprehension targeting Reasoning Over Paragraph Effects in Situations. We target expository language describing causes and effects (e.g., "animal pollinators increase efficiency of fertilization in flowers"), as they have clear implications for new situations. A system is presented a background passage containing at least one of these relations, a novel situation that uses this background, and questions that require reasoning about effects of the relationships in the background passage in the context of the situation. We collect background passages from science textbooks and Wikipedia that contain such phenomena, and ask crowd workers to author situations, questions, and answers, resulting in a 14,322 question dataset. We analyze the challenges of this task and evaluate the performance of state-of-the-art reading comprehension models. The best model performs only slightly better than randomly guessing an answer of the correct type, at 61.6% F1, well below the human performance of 89.0%.

To enable building and testing models on long-document comprehension, we introduce QuALITY, a multiple-choice QA dataset with context passages in English that have an average length of about 5,000 tokens, much longer than typical current models can process. Unlike in prior work with passages, our questions are written and validated by contributors who have read the entire passage, rather than relying on summaries or excerpts. In addition, only half of the questions are answerable by annotators working under tight time constraints, indicating that skimming and simple search are not enough to consistently perform well. Our baseline models perform poorly on this task (55.4%) and significantly lag behind human performance (93.5%).

In-context learning (ICL) allows a language model to improve its problem-solving capability when provided with suitable information in context. Since the choice of in-context information can be determined based on the problem itself, in-context learning is analogous to human learning from teachers in a classroom. Recent works (Didolkar et al., 2024a; 2024b) show that ICL performance can be improved by leveraging a frontier large language model's (LLM) ability to predict required skills to solve a problem, popularly referred to as an LLM's metacognition, and using the recommended skills to construct necessary in-context examples. While this skill-based strategy boosts ICL performance in larger models, its gains on small language models (SLMs) have been minimal, highlighting a performance gap in ICL capabilities. We investigate this gap and show that skill-based prompting can hurt SLM performance on easy questions by introducing unnecessary information, akin to cognitive overload. To address this, we introduce AdaptMI, an adaptive approach to selecting skill-based in-context Math Instructions for SLMs. Inspired by cognitive load theory from human pedagogy, our method only introduces skill-based examples when the model performs poorly. We further propose AdaptMI+, which adds examples targeted to the specific skills missing from the model's responses. On 5-shot evaluations across popular math benchmarks and five SLMs (1B--7B; Qwen, Llama), AdaptMI+ improves accuracy by up to 6% over naive skill-based strategies.

The present study aims to explore the capabilities of Language Models (LMs) in tackling high-stakes multiple-choice tests, represented here by the Exame Nacional do Ensino M\'edio (ENEM), a multidisciplinary entrance examination widely adopted by Brazilian universities. This exam poses challenging tasks for LMs, since its questions may span into multiple fields of knowledge, requiring understanding of information from diverse domains. For instance, a question may require comprehension of both statistics and biology to be solved. This work analyzed responses generated by GPT-3.5 and GPT-4 models for questions presented in the 2009-2017 exams, as well as for questions of the 2022 exam, which were made public after the training of the models was completed. Furthermore, different prompt strategies were tested, including the use of Chain-of-Thought (CoT) prompts to generate explanations for answers. On the 2022 edition, the best-performing model, GPT-4 with CoT, achieved an accuracy of 87%, largely surpassing GPT-3.5 by 11 points. The code and data used on experiments are available at https://github.com/piresramon/gpt-4-enem.

This paper explores the use of large language models (LLMs) as research tools in the history, philosophy, and sociology of science (HPSS). LLMs are remarkably effective at processing unstructured text and inferring meaning from context, offering new affordances that challenge long-standing divides between computational and interpretive methods. This raises both opportunities and challenges for HPSS, which emphasizes interpretive methodologies and understands meaning as context-dependent, ambiguous, and historically situated. We argue that HPSS is uniquely positioned not only to benefit from LLMs' capabilities but also to interrogate their epistemic assumptions and infrastructural implications. To this end, we first offer a concise primer on LLM architectures and training paradigms tailored to non-technical readers. We frame LLMs not as neutral tools but as epistemic infrastructures that encode assumptions about meaning, context, and similarity, conditioned by their training data, architecture, and patterns of use. We then examine how computational techniques enhanced by LLMs, such as structuring data, detecting patterns, and modeling dynamic processes, can be applied to support interpretive research in HPSS. Our analysis compares full-context and generative models, outlines strategies for domain and task adaptation (e.g., continued pretraining, fine-tuning, and retrieval-augmented generation), and evaluates their respective strengths and limitations for interpretive inquiry in HPSS. We conclude with four lessons for integrating LLMs into HPSS: (1) model selection involves interpretive trade-offs; (2) LLM literacy is foundational; (3) HPSS must define its own benchmarks and corpora; and (4) LLMs should enhance, not replace, interpretive methods.

When provided with sufficient explanatory context, smaller Language Models have been shown to exhibit strong reasoning ability on challenging short-answer question-answering tasks where the questions are unseen in training. We evaluate two methods for further improvement in this setting. Both methods focus on combining rationales generated by a larger Language Model with longer contexts created from a multi-hop dense retrieval system. The first method (RR) involves training a Rationale Ranking model to score both generated rationales and retrieved contexts with respect to relevance and truthfulness. We then use the scores to derive combined contexts from both knowledge sources using a number of combinatory strategies. For the second method (RATD) we utilise retrieval-augmented training datasets developed by Hartill et al. 2023 to train a smaller Reasoning model such that it becomes proficient at utilising relevant information from longer text sequences that may be only partially evidential and frequently contain many irrelevant sentences. We find that both methods significantly improve results. Our single best Reasoning model materially improves upon strong comparable prior baselines for unseen evaluation datasets (StrategyQA 58.9 rightarrow 61.7 acc., CommonsenseQA 63.6 rightarrow 72.7 acc., ARC-DA 31.6 rightarrow 52.1 F1, IIRC 25.5 rightarrow 27.3 F1) and a version utilising our prior knowledge of each type of question in selecting a context combination strategy does even better. Our proposed models also generally outperform direct prompts against much larger models (BLOOM 175B and StableVicuna 13B) in both few-shot chain-of-thought and standard few-shot settings.

We measure the performance of in-context learning as a function of task novelty and difficulty for open and closed questions. For that purpose, we created a novel benchmark consisting of hard scientific questions, each paired with a context of various relevancy. We show that counter-intuitively, a context that is more aligned with the topic does not always help more than a less relevant context. This effect is especially visible for open questions and questions of high difficulty or novelty. This result reveals a fundamental difference between the treatment of close-form and open-form questions by large-language models and shows a need for a more robust evaluation of in-context learning on the variety of different types of questions. It also poses a new question of how to optimally select a context for large language models, especially in the context of Retrieval Augmented Generation (RAG) systems. Our results suggest that the answer to this question can be highly application-dependent and might be contingent on factors including the format of the question, the perceived difficulty level of the questions, and the novelty or popularity of the information we seek.

We present a large-scale dataset, ReCoRD, for machine reading comprehension requiring commonsense reasoning. Experiments on this dataset demonstrate that the performance of state-of-the-art MRC systems fall far behind human performance. ReCoRD represents a challenge for future research to bridge the gap between human and machine commonsense reading comprehension. ReCoRD is available at http://nlp.jhu.edu/record.

We describe a Question Answering (QA) dataset that contains complex questions with conditional answers, i.e. the answers are only applicable when certain conditions apply. We call this dataset ConditionalQA. In addition to conditional answers, the dataset also features: (1) long context documents with information that is related in logically complex ways; (2) multi-hop questions that require compositional logical reasoning; (3) a combination of extractive questions, yes/no questions, questions with multiple answers, and not-answerable questions; (4) questions asked without knowing the answers. We show that ConditionalQA is challenging for many of the existing QA models, especially in selecting answer conditions. We believe that this dataset will motivate further research in answering complex questions over long documents. Data and leaderboard are publicly available at https://github.com/haitian-sun/ConditionalQA.

In-context learning (ICL), the ability of large language models to perform novel tasks by conditioning on a prompt with a few task examples, requires these examples to be informative about the test instance. The standard approach of independently ranking and selecting the most similar examples selects redundant examples while omitting important information. In this work, we show that BERTScore-Recall (BSR) selects better examples that demonstrate more of the salient aspects, e.g. reasoning patterns, of the test input. We further extend BSR and many standard metrics to easily optimizable set-level metrics, giving still better coverage of those salient aspects. On 15 datasets spanning 6 tasks and with 7 diverse LLMs, we show that (1) BSR is the superior metric for in-context example selection across the board, and (2) for compositional tasks, set selection using Set-BSR outperforms independent ranking by up to 17 points on average and, despite being training-free, surpasses methods that leverage task or LLM-specific training.

Despite recent advancements in Large Language Models (LLMs), their performance on tasks involving long contexts remains sub-optimal. In-Context Learning (ICL) with few-shot examples may be an appealing solution to enhance LLM performance in this scenario; However, naively adding ICL examples with long context introduces challenges, including substantial token overhead added for each few-shot example and context mismatch between the demonstrations and the target query. In this work, we propose to automatically generate few-shot examples for long context QA tasks by recycling contexts. Specifically, given a long input context (1-3k tokens) and a query, we generate additional query-output pairs from the given context as few-shot examples, while introducing the context only once. This ensures that the demonstrations are leveraging the same context as the target query while only adding a small number of tokens to the prompt. We further enhance each demonstration by instructing the model to explicitly identify the relevant paragraphs before the answer, which improves performance while providing fine-grained attribution to the answer source. We apply our method on multiple LLMs and obtain substantial improvements (+23\% on average across models) on various QA datasets with long context, especially when the answer lies within the middle of the context. Surprisingly, despite introducing only single-hop ICL examples, LLMs also successfully generalize to multi-hop long-context QA using our approach.

Designing high-quality educational questions is a challenging and time-consuming task. In this work, we propose a novel approach that utilizes prompt-based techniques to generate descriptive and reasoning-based questions. However, current question-answering (QA) datasets are inadequate for conducting our experiments on prompt-based question generation (QG) in an educational setting. Therefore, we curate a new QG dataset called EduProbe for school-level subjects, by leveraging the rich content of NCERT textbooks. We carefully annotate this dataset as quadruples of 1) Context: a segment upon which the question is formed; 2) Long Prompt: a long textual cue for the question (i.e., a longer sequence of words or phrases, covering the main theme of the context); 3) Short Prompt: a short textual cue for the question (i.e., a condensed representation of the key information or focus of the context); 4) Question: a deep question that aligns with the context and is coherent with the prompts. We investigate several prompt-based QG methods by fine-tuning pre-trained transformer-based large language models (LLMs), namely PEGASUS, T5, MBART, and BART. Moreover, we explore the performance of two general-purpose pre-trained LLMs such as Text-Davinci-003 and GPT-3.5-Turbo without any further training. By performing automatic evaluation, we show that T5 (with long prompt) outperforms all other models, but still falls short of the human baseline. Under human evaluation criteria, TextDavinci-003 usually shows better results than other models under various prompt settings. Even in the case of human evaluation criteria, QG models mostly fall short of the human baseline. Our code and dataset are available at: https://github.com/my625/PromptQG

Large language models have demonstrated impressive performance on challenging mathematical reasoning tasks, which has triggered the discussion of whether the performance is achieved by true reasoning capability or memorization. To investigate this question, prior work has constructed mathematical benchmarks when questions undergo simple perturbations -- modifications that still preserve the underlying reasoning patterns of the solutions. However, no work has explored hard perturbations, which fundamentally change the nature of the problem so that the original solution steps do not apply. To bridge the gap, we construct MATH-P-Simple and MATH-P-Hard via simple perturbation and hard perturbation, respectively. Each consists of 279 perturbed math problems derived from level-5 (hardest) problems in the MATH dataset (Hendrycksmath et. al., 2021). We observe significant performance drops on MATH-P-Hard across various models, including o1-mini (-16.49%) and gemini-2.0-flash-thinking (-12.9%). We also raise concerns about a novel form of memorization where models blindly apply learned problem-solving skills without assessing their applicability to modified contexts. This issue is amplified when using original problems for in-context learning. We call for research efforts to address this challenge, which is critical for developing more robust and reliable reasoning models.

We describe new datasets for studying generalization from easy to hard examples.

The large language model (LLM)-as-judge paradigm has been used to meet the demand for a cheap, reliable, and fast evaluation of model outputs during AI system development and post-deployment monitoring. While judge models -- LLMs finetuned to specialize in assessing and critiquing model outputs -- have been touted as general purpose evaluators, they are typically evaluated only on non-contextual scenarios, such as instruction following. The omission of contextual settings -- those where external information is used as context to generate an output -- is surprising given the increasing prevalence of retrieval-augmented generation (RAG) and summarization use cases. Contextual assessment is uniquely challenging, as evaluation often depends on practitioner priorities, leading to conditional evaluation criteria (e.g., comparing responses based on factuality and then considering completeness if they are equally factual). To address the gap, we propose ContextualJudgeBench, a judge benchmark with 2,000 challenging response pairs across eight splits inspired by real-world contextual evaluation scenarios. We build our benchmark with a multi-pronged data construction pipeline that leverages both existing human annotations and model-based perturbations. Our comprehensive study across 11 judge models and 9 general purpose models, reveals that the contextual information and its assessment criteria present a significant challenge to even state-of-the-art models. For example, OpenAI's o1, the best-performing model, barely reaches 55% consistent accuracy.

The abilities of long-context language models (LMs) are often evaluated using the "Needle-in-a-Haystack" (NIAH) test, which comprises tasks designed to assess a model's ability to identify specific information ("needle") within large text sequences ("haystack"). While these benchmarks measure how well models understand long-context input sequences, they do not effectively gauge the quality of long-form text generation--a critical aspect for applications such as design proposals and creative writing. To address this gap, we have introduced a new long-form text evaluation benchmark, Spinning the Golden Thread (SGT), which tests models' ability to identify specific events within generated long text sequences. In this benchmark, we prompt long-context LMs to create long-form text that must include particular events or constraints and evaluate their ability to incorporate these elements. We evaluated ten long-context LMs across four distinct scenarios, three types of prompt instructions, and two different generation-length settings (16K and 32K). Although these models perform well on NIAH benchmarks, none demonstrated satisfactory performance on the Spinning the Golden Thread, raising concerns about their ability to generate coherent long-form text that follows instructions. Additionally, as the length of the generated text increases, all models exhibit a significant drop in performance.

We introduce FACTS Grounding, an online leaderboard and associated benchmark that evaluates language models' ability to generate text that is factually accurate with respect to given context in the user prompt. In our benchmark, each prompt includes a user request and a full document, with a maximum length of 32k tokens, requiring long-form responses. The long-form responses are required to be fully grounded in the provided context document while fulfilling the user request. Models are evaluated using automated judge models in two phases: (1) responses are disqualified if they do not fulfill the user request; (2) they are judged as accurate if the response is fully grounded in the provided document. The automated judge models were comprehensively evaluated against a held-out test-set to pick the best prompt template, and the final factuality score is an aggregate of multiple judge models to mitigate evaluation bias. The FACTS Grounding leaderboard will be actively maintained over time, and contains both public and private splits to allow for external participation while guarding the integrity of the leaderboard. It can be found at https://www.kaggle.com/facts-leaderboard.

Recent advancements in long-context Large Language Models (LLMs) have primarily concentrated on processing extended input contexts, resulting in significant strides in long-context comprehension. However, the equally critical aspect of generating long-form outputs has received comparatively less attention. This paper advocates for a paradigm shift in NLP research toward addressing the challenges of long-output generation. Tasks such as novel writing, long-term planning, and complex reasoning require models to understand extensive contexts and produce coherent, contextually rich, and logically consistent extended text. These demands highlight a critical gap in current LLM capabilities. We underscore the importance of this under-explored domain and call for focused efforts to develop foundational LLMs tailored for generating high-quality, long-form outputs, which hold immense potential for real-world applications.

When people answer questions about a specific situation, e.g., "I cheated on my mid-term exam last week. Was that wrong?", cognitive science suggests that they form a mental picture of that situation before answering. While we do not know how language models (LMs) answer such questions, we conjecture that they may answer more accurately if they are also provided with additional details about the question situation, elaborating the "scene". To test this conjecture, we train a new model, DREAM, to answer questions that elaborate the scenes that situated questions are about, and then provide those elaborations as additional context to a question-answering (QA) model. We find that DREAM is able to create better scene elaborations (more accurate, useful, and consistent) than a representative state-of-the-art, zero-shot model (Macaw). We also find that using the scene elaborations as additional context improves the answer accuracy of a downstream QA system, including beyond that obtainable by simply further finetuning the QA system on DREAM's training data. These results suggest that adding focused elaborations about a situation can improve a system's reasoning about it, and may serve as an effective way of injecting new scenario based knowledge into QA models. Finally, our approach is dataset-neutral; we observe improved QA performance across different models, with even bigger gains on models with fewer parameters. We make our dataset and model publicly available at https://github.com/allenai/dream.

Natural language understanding (NLU) is integral to various social media applications. However, existing NLU models rely heavily on context for semantic learning, resulting in compromised performance when faced with short and noisy social media content. To address this issue, we leverage in-context learning (ICL), wherein language models learn to make inferences by conditioning on a handful of demonstrations to enrich the context and propose a novel hashtag-driven in-context learning (HICL) framework. Concretely, we pre-train a model #Encoder, which employs #hashtags (user-annotated topic labels) to drive BERT-based pre-training through contrastive learning. Our objective here is to enable #Encoder to gain the ability to incorporate topic-related semantic information, which allows it to retrieve topic-related posts to enrich contexts and enhance social media NLU with noisy contexts. To further integrate the retrieved context with the source text, we employ a gradient-based method to identify trigger terms useful in fusing information from both sources. For empirical studies, we collected 45M tweets to set up an in-context NLU benchmark, and the experimental results on seven downstream tasks show that HICL substantially advances the previous state-of-the-art results. Furthermore, we conducted extensive analyzes and found that: (1) combining source input with a top-retrieved post from #Encoder is more effective than using semantically similar posts; (2) trigger words can largely benefit in merging context from the source and retrieved posts.

The Needle-in-a-Haystack (NIAH) benchmark is widely used to evaluate Large Language Models' (LLMs) ability to understand long contexts (LC). It evaluates the capability to identify query-relevant context within extensive query-irrelevant passages. Although this method serves as a widely accepted standard for evaluating long-context understanding, our findings suggest it may overestimate the true LC capability of LLMs. We demonstrate that even state-of-the-art models such as GPT-4o struggle to intactly incorporate given contexts made up of solely query-relevant ten sentences. In response, we introduce a novel benchmark, NeedleChain, where the context consists entirely of query-relevant information, requiring the LLM to fully grasp the input to answer correctly. Our benchmark allows for flexible context length and reasoning order, offering a more comprehensive analysis of LLM performance. Additionally, we propose an extremely simple yet compelling strategy to improve LC understanding capability of LLM: ROPE Contraction. Our experiments with various advanced LLMs reveal a notable disparity between their ability to process large contexts and their capacity to fully understand them. Source code and datasets are available at https://github.com/hyeonseokk/NeedleChain

Recently, reading comprehension models achieved near-human performance on large-scale datasets such as SQuAD, CoQA, MS Macro, RACE, etc. This is largely due to the release of pre-trained contextualized representations such as BERT and ELMo, which can be fine-tuned for the target task. Despite those advances and the creation of more challenging datasets, most of the work is still done for English. Here, we study the effectiveness of multilingual BERT fine-tuned on large-scale English datasets for reading comprehension (e.g., for RACE), and we apply it to Bulgarian multiple-choice reading comprehension. We propose a new dataset containing 2,221 questions from matriculation exams for twelfth grade in various subjects -history, biology, geography and philosophy-, and 412 additional questions from online quizzes in history. While the quiz authors gave no relevant context, we incorporate knowledge from Wikipedia, retrieving documents matching the combination of question + each answer option. Moreover, we experiment with different indexing and pre-training strategies. The evaluation results show accuracy of 42.23%, which is well above the baseline of 24.89%.

Our work presents a novel angle for evaluating language models' (LMs) mathematical abilities, by investigating whether they can discern skills and concepts enabled by math content. We contribute two datasets: one consisting of 385 fine-grained descriptions of K-12 math skills and concepts, or standards, from Achieve the Core (ATC), and another of 9.9K problems labeled with these standards (MathFish). Working with experienced teachers, we find that LMs struggle to tag and verify standards linked to problems, and instead predict labels that are close to ground truth, but differ in subtle ways. We also show that LMs often generate problems that do not fully align with standards described in prompts. Finally, we categorize problems in GSM8k using math standards, allowing us to better understand why some problems are more difficult to solve for models than others.

The remarkable capability of large language models (LLMs) for in-context learning (ICL) needs to be activated by demonstration examples. Prior work has extensively explored the selection of examples for ICL, predominantly following the "select then organize" paradigm, such approaches often neglect the internal relationships between examples and exist an inconsistency between the training and inference. In this paper, we formulate the problem as a sequential selection problem and introduce Se^2, a sequential-aware method that leverages the LLM's feedback on varying context, aiding in capturing inter-relationships and sequential information among examples, significantly enriching the contextuality and relevance of ICL prompts. Meanwhile, we utilize beam search to seek and construct example sequences, enhancing both quality and diversity. Extensive experiments across 23 NLP tasks from 8 distinct categories illustrate that Se^2 markedly surpasses competitive baselines and achieves 42% relative improvement over random selection. Further in-depth analysis show the effectiveness of proposed strategies, highlighting Se^2's exceptional stability and adaptability across various scenarios. Our code will be released to facilitate future research.

Long-context language models (LCLM), characterized by their extensive context window, is becoming increasingly popular. Meanwhile, many long-context benchmarks present challenging tasks that even the most advanced LCLMs struggle to complete. However, the underlying sources of various challenging long-context tasks have seldom been studied. To bridge this gap, we conduct experiments to indicate their difficulty stems primarily from two basic issues: "multi-matching retrieval," which requires the simultaneous retrieval of multiple items, and "logic-based retrieval," which necessitates logical judgment within retrieval criteria. These two problems, while seemingly straightforward, actually exceed the capabilities of LCLMs because they are proven to be hyper-multi-step (demanding numerous steps to solve) in nature. This finding could explain why LLMs struggle with more advanced long-context tasks, providing a more accurate perspective for rethinking solutions for them.

Teachers and students are increasingly relying on online learning resources to supplement the ones provided in school. This increase in the breadth and depth of available resources is a great thing for students, but only provided they are able to find answers to their queries. Question-answering and information retrieval systems have benefited from public datasets to train and evaluate their algorithms, but most of these datasets have been in English text written by and for adults. We introduce a new public French question-answering dataset collected from Alloprof, a Quebec-based primary and high-school help website, containing 29 349 questions and their explanations in a variety of school subjects from 10 368 students, with more than half of the explanations containing links to other questions or some of the 2 596 reference pages on the website. We also present a case study of this dataset in an information retrieval task. This dataset was collected on the Alloprof public forum, with all questions verified for their appropriateness and the explanations verified both for their appropriateness and their relevance to the question. To predict relevant documents, architectures using pre-trained BERT models were fine-tuned and evaluated. This dataset will allow researchers to develop question-answering, information retrieval and other algorithms specifically for the French speaking education context. Furthermore, the range of language proficiency, images, mathematical symbols and spelling mistakes will necessitate algorithms based on a multimodal comprehension. The case study we present as a baseline shows an approach that relies on recent techniques provides an acceptable performance level, but more work is necessary before it can reliably be used and trusted in a production setting.

Employing Large Language Models (LLMs) to address mathematical problems is an intriguing research endeavor, considering the abundance of math problems expressed in natural language across numerous science and engineering fields. While several prior works have investigated solving elementary mathematics using LLMs, this work explores the frontier of using GPT-4 for solving more complex and challenging math problems. We evaluate various ways of using GPT-4. Some of them are adapted from existing work, and one is \MathChat, a conversational problem-solving framework newly proposed in this work. We perform the evaluation on difficult high school competition problems from the MATH dataset, which shows the advantage of the proposed conversational approach.

We propose EXAMS -- a new benchmark dataset for cross-lingual and multilingual question answering for high school examinations. We collected more than 24,000 high-quality high school exam questions in 16 languages, covering 8 language families and 24 school subjects from Natural Sciences and Social Sciences, among others. EXAMS offers a fine-grained evaluation framework across multiple languages and subjects, which allows precise analysis and comparison of various models. We perform various experiments with existing top-performing multilingual pre-trained models and we show that EXAMS offers multiple challenges that require multilingual knowledge and reasoning in multiple domains. We hope that EXAMS will enable researchers to explore challenging reasoning and knowledge transfer methods and pre-trained models for school question answering in various languages which was not possible before. The data, code, pre-trained models, and evaluation are available at https://github.com/mhardalov/exams-qa.

Math Word Problems (MWPs) play a vital role in assessing the capabilities of Large Language Models (LLMs), yet current research primarily focuses on questions with concise contexts. The impact of longer contexts on mathematical reasoning remains under-explored. This study pioneers the investigation of Context Length Generalizability (CoLeG), which refers to the ability of LLMs to solve MWPs with extended narratives. We introduce Extended Grade-School Math (E-GSM), a collection of MWPs featuring lengthy narratives, and propose two novel metrics to evaluate the efficacy and resilience of LLMs in tackling these problems. Our analysis of existing zero-shot prompting techniques with proprietary LLMs along with open-source LLMs reveals a general deficiency in CoLeG. To alleviate these issues, we propose tailored approaches for different categories of LLMs. For proprietary LLMs, we introduce a new instructional prompt designed to mitigate the impact of long contexts. For open-source LLMs, we develop a novel auxiliary task for fine-tuning to enhance CoLeG. Our comprehensive results demonstrate the effectiveness of our proposed methods, showing improved performance on E-GSM. Additionally, we conduct an in-depth analysis to differentiate the effects of semantic understanding and reasoning efficacy, showing that our methods improves the latter. We also establish the generalizability of our methods across several other MWP benchmarks. Our findings highlight the limitations of current LLMs and offer practical solutions correspondingly, paving the way for further exploration of model generalizability and training methodologies.

We introduce a high-quality dataset that contains 3,397 samples comprising (i) multiple choice questions, (ii) answers (including distractors), and (iii) their source documents, from the educational domain. Each question is phrased in two forms, normal and close. Correct answers are linked to source documents with sentence-level annotations. Thus, our versatile dataset can be used for both question and distractor generation, as well as to explore new challenges such as question format conversion. Furthermore, 903 questions are accompanied by their cognitive complexity level as per Bloom's taxonomy. All questions have been generated by educational experts rather than crowd workers to ensure they are maintaining educational and learning standards. Our analysis and experiments suggest distinguishable differences between our dataset and commonly used ones for question generation for educational purposes. We believe this new dataset can serve as a valuable resource for research and evaluation in the educational domain. The dataset and baselines will be released to support further research in question generation.

Large language model (LLM) providers boast big numbers for maximum context window sizes. To test the real world use of context windows, we 1) define a concept of maximum effective context window, 2) formulate a testing method of a context window's effectiveness over various sizes and problem types, and 3) create a standardized way to compare model efficacy for increasingly larger context window sizes to find the point of failure. We collected hundreds of thousands of data points across several models and found significant differences between reported Maximum Context Window (MCW) size and Maximum Effective Context Window (MECW) size. Our findings show that the MECW is, not only, drastically different from the MCW but also shifts based on the problem type. A few top of the line models in our test group failed with as little as 100 tokens in context; most had severe degradation in accuracy by 1000 tokens in context. All models fell far short of their Maximum Context Window by as much as 99 percent. Our data reveals the Maximum Effective Context Window shifts based on the type of problem provided, offering clear and actionable insights into how to improve model accuracy and decrease model hallucination rates.

We propose a new test to measure a text model's multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more. To attain high accuracy on this test, models must possess extensive world knowledge and problem solving ability. We find that while most recent models have near random-chance accuracy, the very largest GPT-3 model improves over random chance by almost 20 percentage points on average. However, on every one of the 57 tasks, the best models still need substantial improvements before they can reach expert-level accuracy. Models also have lopsided performance and frequently do not know when they are wrong. Worse, they still have near-random accuracy on some socially important subjects such as morality and law. By comprehensively evaluating the breadth and depth of a model's academic and professional understanding, our test can be used to analyze models across many tasks and to identify important shortcomings.

In-context Learning (ICL) enables large language models (LLMs) to tackle downstream tasks through sophisticated prompting and high-quality demonstrations. However, this traditional ICL paradigm shows limitations when facing complex mathematical reasoning tasks, primarily due to its heavy dependence on example quality and the necessity for human intervention in challenging scenarios. To address these limitations, this paper presents HiAR-ICL, a High-level Automated Reasoning paradigm in ICL that shifts focus from specific examples to abstract thinking patterns, extending the conventional concept of context in ICL. HiAR-ICL introduces five atomic reasoning actions as fundamental components for constructing chain-structured patterns. Using Monte Carlo Tree Search, we explore reasoning paths and construct thought cards to guide subsequent inference. We then develop a cognitive complexity framework that dynamically matches problems with appropriate thought cards. Experimental results demonstrate HiAR-ICL's effectiveness, achieving state-of-the-art accuracy (79.6%) on the MATH benchmark with Qwen2.5-7B-Instruct, surpassing GPT-4o (76.6%) and Claude 3.5 (71.1%).

Reading is integral to everyday life, and yet learning to read is a struggle for many young learners. During lessons, teachers can use comprehension questions to increase engagement, test reading skills, and improve retention. Historically such questions were written by skilled teachers, but recently language models have been used to generate comprehension questions. However, many existing Question Generation (QG) systems focus on generating literal questions from the text, and have no way to control the type of the generated question. In this paper, we study QG for reading comprehension where inferential questions are critical and extractive techniques cannot be used. We propose a two-step model (HTA-WTA) that takes advantage of previous datasets, and can generate questions for a specific targeted comprehension skill. We propose a new reading comprehension dataset that contains questions annotated with story-based reading comprehension skills (SBRCS), allowing for a more complete reader assessment. Across several experiments, our results show that HTA-WTA outperforms multiple strong baselines on this new dataset. We show that the HTA-WTA model tests for strong SCRS by asking deep inferential questions.

Large language models (LLM) have recently shown the extraordinary ability to perform unseen tasks based on few-shot examples provided as text, also known as in-context learning (ICL). While recent works have attempted to understand the mechanisms driving ICL, few have explored training strategies that incentivize these models to generalize to multiple tasks. Multi-task learning (MTL) for generalist models is a promising direction that offers transfer learning potential, enabling large parameterized models to be trained from simpler, related tasks. In this work, we investigate the combination of MTL with ICL to build models that efficiently learn tasks while being robust to out-of-distribution examples. We propose several effective curriculum learning strategies that allow ICL models to achieve higher data efficiency and more stable convergence. Our experiments reveal that ICL models can effectively learn difficult tasks by training on progressively harder tasks while mixing in prior tasks, denoted as mixed curriculum in this work. Our code and models are available at https://github.com/harmonbhasin/curriculum_learning_icl .

We introduce a new challenge to test the STEM skills of neural models. The problems in the real world often require solutions, combining knowledge from STEM (science, technology, engineering, and math). Unlike existing datasets, our dataset requires the understanding of multimodal vision-language information of STEM. Our dataset features one of the largest and most comprehensive datasets for the challenge. It includes 448 skills and 1,073,146 questions spanning all STEM subjects. Compared to existing datasets that often focus on examining expert-level ability, our dataset includes fundamental skills and questions designed based on the K-12 curriculum. We also add state-of-the-art foundation models such as CLIP and GPT-3.5-Turbo to our benchmark. Results show that the recent model advances only help master a very limited number of lower grade-level skills (2.5% in the third grade) in our dataset. In fact, these models are still well below (averaging 54.7%) the performance of elementary students, not to mention near expert-level performance. To understand and increase the performance on our dataset, we teach the models on a training split of our dataset. Even though we observe improved performance, the model performance remains relatively low compared to average elementary students. To solve STEM problems, we will need novel algorithmic innovations from the community.

Online toxic language causes real harm, especially in regions with limited moderation tools. In this study, we evaluate how large language models handle toxic comments in Serbian, Croatian, and Bosnian, languages with limited labeled data. We built and manually labeled a dataset of 4,500 YouTube and TikTok comments drawn from videos across diverse categories, including music, politics, sports, modeling, influencer content, discussions of sexism, and general topics. Four models (GPT-3.5 Turbo, GPT-4.1, Gemini 1.5 Pro, and Claude 3 Opus) were tested in two modes: zero-shot and context-augmented. We measured precision, recall, F1 score, accuracy and false positive rates. Including a short context snippet raised recall by about 0.12 on average and improved F1 score by up to 0.10, though it sometimes increased false positives. The best balance came from Gemini in context-augmented mode, reaching an F1 score of 0.82 and accuracy of 0.82, while zero-shot GPT-4.1 led on precision and had the lowest false alarms. We show how adding minimal context can improve toxic language detection in low-resource settings and suggest practical strategies such as improved prompt design and threshold calibration. These results show that prompt design alone can yield meaningful gains in toxicity detection for underserved Balkan language communities.

Large Language Models (LLMs) demonstrate partial forecasting competence across social, political, and economic events. Yet, their predictive ability varies sharply with domain structure and prompt framing. We investigate how forecasting performance varies with different model families on real-world questions about events that happened beyond the model cutoff date. We analyze how context, question type, and external knowledge affect accuracy and calibration, and how adding factual news context modifies belief formation and failure modes. Our results show that forecasting ability is highly variable as it depends on what, and how, we ask.

Large language models (LLMs) are increasingly integrated into applications ranging from review summarization to medical diagnosis support, where they affect human decisions. Even though LLMs perform well in many tasks, they may also inherit societal or cognitive biases, which can inadvertently transfer to humans. We investigate when and how LLMs expose users to biased content and quantify its severity. Specifically, we assess three LLM families in summarization and news fact-checking tasks, evaluating how much LLMs stay consistent with their context and/or hallucinate. Our findings show that LLMs expose users to content that changes the sentiment of the context in 21.86% of the cases, hallucinates on post-knowledge-cutoff data questions in 57.33% of the cases, and primacy bias in 5.94% of the cases. We evaluate 18 distinct mitigation methods across three LLM families and find that targeted interventions can be effective. Given the prevalent use of LLMs in high-stakes domains, such as healthcare or legal analysis, our results highlight the need for robust technical safeguards and for developing user-centered interventions that address LLM limitations.

We extend Textual Inversion to learn pseudo-words that represent a concept at different resolutions. This allows us to generate images that use the concept with different levels of detail and also to manipulate different resolutions using language. Once learned, the user can generate images at different levels of agreement to the original concept; "A photo of S^*(0)" produces the exact object while the prompt "A photo of S^*(0.8)" only matches the rough outlines and colors. Our framework allows us to generate images that use different resolutions of an image (e.g. details, textures, styles) as separate pseudo-words that can be composed in various ways. We open-soure our code in the following URL: https://github.com/giannisdaras/multires_textual_inversion

We propose MMLU-SR, a novel dataset designed to measure the true comprehension abilities of Large Language Models (LLMs) by challenging their performance in question-answering tasks with modified terms. We reasoned that an agent that ``truly'' understands a concept can still evaluate it when key terms are replaced by suitably defined alternate terms, and sought to differentiate such comprehension from mere text replacement. In our study, we modified standardized test questions by replacing a key term with a dummy word along with its definition. The key term could be in the context of questions, answers, or both questions and answers. Notwithstanding the high scores achieved by recent popular LLMs on the MMLU leaderboard, we found a substantial reduction in model performance after such replacement, suggesting poor comprehension. This new benchmark provides a rigorous benchmark for testing true model comprehension, and poses a challenge to the broader scientific community.

Large language models (LMs) such as GPT-3 have the surprising ability to do in-context learning, where the model learns to do a downstream task simply by conditioning on a prompt consisting of input-output examples. The LM learns from these examples without being explicitly pretrained to learn. Thus, it is unclear what enables in-context learning. In this paper, we study how in-context learning can emerge when pretraining documents have long-range coherence. Here, the LM must infer a latent document-level concept to generate coherent next tokens during pretraining. At test time, in-context learning occurs when the LM also infers a shared latent concept between examples in a prompt. We prove when this occurs despite a distribution mismatch between prompts and pretraining data in a setting where the pretraining distribution is a mixture of HMMs. In contrast to messy large-scale datasets used to train LMs capable of in-context learning, we generate a small-scale synthetic dataset (GINC) where Transformers and LSTMs both exhibit in-context learning. Beyond the theory, experiments on GINC exhibit large-scale real-world phenomena including improved in-context performance with model scaling (despite the same pretraining loss), sensitivity to example order, and instances where zero-shot is better than few-shot in-context learning.

Existing dense retrieval models struggle with reasoning-intensive retrieval task as they fail to capture implicit relevance that requires reasoning beyond surface-level semantic information. To address these challenges, we propose Scenario-Profiled Indexing with Knowledge Expansion (SPIKE), a dense retrieval framework that explicitly indexes implicit relevance by decomposing documents into scenario-based retrieval units. SPIKE organizes documents into scenario, which encapsulates the reasoning process necessary to uncover implicit relationships between hypothetical information needs and document content. SPIKE constructs a scenario-augmented dataset using a powerful teacher large language model (LLM), then distills these reasoning capabilities into a smaller, efficient scenario generator. During inference, SPIKE incorporates scenario-level relevance alongside document-level relevance, enabling reasoning-aware retrieval. Extensive experiments demonstrate that SPIKE consistently enhances retrieval performance across various query types and dense retrievers. It also enhances the retrieval experience for users through scenario and offers valuable contextual information for LLMs in retrieval-augmented generation (RAG).

Prompting Large Language Models (LLMs), or providing context on the expected model of operation, is an effective way to steer the outputs of such models to satisfy human desiderata after they have been trained. But in rapidly evolving domains, there is often need to fine-tune LLMs to improve either the kind of knowledge in their memory or their abilities to perform open ended reasoning in new domains. When human's learn new concepts, we often do so by linking the new material that we are studying to concepts we have already learned before. To that end, we ask, "can prompting help us teach LLMs how to learn". In this work, we study a novel generalization of instruction tuning, called contextual fine-tuning, to fine-tune LLMs. Our method leverages instructional prompts designed to mimic human cognitive strategies in learning and problem-solving to guide the learning process during training, aiming to improve the model's interpretation and understanding of domain-specific knowledge. We empirically demonstrate that this simple yet effective modification improves the ability of LLMs to be fine-tuned rapidly on new datasets both within the medical and financial domains.

Leveraging Large Language Models (LLMs) as judges for evaluating the performance of LLMs has recently garnered attention. Nonetheless, this type of approach concurrently introduces potential biases from LLMs, raising concerns about the reliability of the evaluation results. To mitigate this issue, we propose and study two versions of many-shot in-context prompts, Reinforced and Unsupervised ICL, for helping GPT-4o-as-a-Judge in single answer grading. The former uses in-context examples with model-generated rationales, and the latter without. Based on the designed prompts, we investigate the impact of scaling the number of in-context examples on the agreement and quality of the evaluation. Furthermore, we first reveal the symbol bias in GPT-4o-as-a-Judge for pairwise comparison and then propose a simple yet effective approach to mitigate it. Experimental results show that advanced long-context LLMs, such as GPT-4o, perform better in the many-shot regime than in the zero-shot regime. Meanwhile, the experimental results further verify the effectiveness of the symbol bias mitigation approach.

Large language models have achieved impressive performance on various natural language processing tasks. However, so far they have been evaluated primarily on benchmarks where all information in the input context is relevant for solving the task. In this work, we investigate the distractibility of large language models, i.e., how the model problem-solving accuracy can be influenced by irrelevant context. In particular, we introduce Grade-School Math with Irrelevant Context (GSM-IC), an arithmetic reasoning dataset with irrelevant information in the problem description. We use this benchmark to measure the distractibility of cutting-edge prompting techniques for large language models, and find that the model performance is dramatically decreased when irrelevant information is included. We also identify several approaches for mitigating this deficiency, such as decoding with self-consistency and adding to the prompt an instruction that tells the language model to ignore the irrelevant information.

Recent advances in large language models (LLMs), such as ChatGPT, have led to highly sophisticated conversation agents. However, these models suffer from "hallucinations," where the model generates false or fabricated information. Addressing this challenge is crucial, particularly with AI-driven platforms being adopted across various sectors. In this paper, we propose a novel method to recognize and flag instances when LLMs perform outside their domain knowledge, and ensuring users receive accurate information. We find that the use of context combined with embedded tags can successfully combat hallucinations within generative language models. To do this, we baseline hallucination frequency in no-context prompt-response pairs using generated URLs as easily-tested indicators of fabricated data. We observed a significant reduction in overall hallucination when context was supplied along with question prompts for tested generative engines. Lastly, we evaluated how placing tags within contexts impacted model responses and were able to eliminate hallucinations in responses with 98.88% effectiveness.

Improvements in language models' capabilities have pushed their applications towards longer contexts, making long-context evaluation and development an active research area. However, many disparate use-cases are grouped together under the umbrella term of "long-context", defined simply by the total length of the model's input, including - for example - Needle-in-a-Haystack tasks, book summarization, and information aggregation. Given their varied difficulty, in this position paper we argue that conflating different tasks by their context length is unproductive. As a community, we require a more precise vocabulary to understand what makes long-context tasks similar or different. We propose to unpack the taxonomy of long-context based on the properties that make them more difficult with longer contexts. We propose two orthogonal axes of difficulty: (I) Diffusion: How hard is it to find the necessary information in the context? (II) Scope: How much necessary information is there to find? We survey the literature on long-context, provide justification for this taxonomy as an informative descriptor, and situate the literature with respect to it. We conclude that the most difficult and interesting settings, whose necessary information is very long and highly diffused within the input, is severely under-explored. By using a descriptive vocabulary and discussing the relevant properties of difficulty in long-context, we can implement more informed research in this area. We call for a careful design of tasks and benchmarks with distinctly long context, taking into account the characteristics that make it qualitatively different from shorter context.

The web provides a rich, open-domain environment with textual, structural, and spatial properties. We propose a new task for grounding language in this environment: given a natural language command (e.g., "click on the second article"), choose the correct element on the web page (e.g., a hyperlink or text box). We collected a dataset of over 50,000 commands that capture various phenomena such as functional references (e.g. "find who made this site"), relational reasoning (e.g. "article by john"), and visual reasoning (e.g. "top-most article"). We also implemented and analyzed three baseline models that capture different phenomena present in the dataset.

In-context learning (ICL) is a powerful paradigm emerged from large language models (LLMs). Despite its promises, ICL performance is known to be highly sensitive to input examples. In this work, we use in-context influences to analyze few-shot ICL performance directly from the in-context examples. Our proposed influence-based example selection method can identify both positive and negative examples, outperforming several baselines when evaluated on 9 SuperGLUE tasks. Our analysis uncovers up to a 16.3% performance gap between using the most negative in-context examples compared to the most positive. In a case study, we apply our influence-based framework to quantify the phenomena of recency bias in example ordering for few-shot ICL.

Context lengths for models have grown rapidly, from thousands to millions of tokens in just a few years. The extreme context sizes of modern long-context models have made it difficult to construct realistic long-context benchmarks -- not only due to the cost of collecting million-context tasks but also in identifying realistic scenarios that require significant contexts. We identify code comprehension and repair as a natural testbed and challenge task for long-context models and introduce LongCodeBench (LCB), a benchmark to test LLM coding abilities in long-context scenarios. Our benchmark tests both the comprehension and repair capabilities of LCLMs in realistic and important settings by drawing from real-world GitHub issues and constructing QA (LongCodeQA) and bug fixing (LongSWE-Bench) tasks. We carefully stratify the complexity of our benchmark, enabling us to evaluate models across different scales -- ranging from Qwen2.5 14B Instruct to Google's flagship Gemini model. We find that long-context remains a weakness for all models, with performance drops such as from 29% to 3% for Claude 3.5 Sonnet, or from 70.2% to 40% for Qwen2.5.

Recently, the physical capabilities of (M)LLMs have garnered increasing attention. However, existing benchmarks for physics suffer from two major gaps: they neither provide systematic and up-to-date coverage of real-world physics competitions such as physics Olympiads, nor enable direct performance comparison with humans. To bridge these gaps, we present HiPhO, the first benchmark dedicated to high school physics Olympiads with human-aligned evaluation. Specifically, HiPhO highlights three key innovations. (1) Comprehensive Data: It compiles 13 latest Olympiad exams from 2024-2025, spanning both international and regional competitions, and covering mixed modalities that encompass problems spanning text-only to diagram-based. (2) Professional Evaluation: We adopt official marking schemes to perform fine-grained grading at both the answer and step level, fully aligned with human examiners to ensure high-quality and domain-specific evaluation. (3) Comparison with Human Contestants: We assign gold, silver, and bronze medals to models based on official medal thresholds, thereby enabling direct comparison between (M)LLMs and human contestants. Our large-scale evaluation of 30 state-of-the-art (M)LLMs shows that: across 13 exams, open-source MLLMs mostly remain at or below the bronze level; open-source LLMs show promising progress with occasional golds; closed-source reasoning MLLMs can achieve 6 to 12 gold medals; and most models still have a significant gap from full marks. These results highlight a substantial performance gap between open-source models and top students, the strong physical reasoning capabilities of closed-source reasoning models, and the fact that there is still significant room for improvement. HiPhO, as a rigorous, human-aligned, and Olympiad-focused benchmark for advancing multimodal physical reasoning, is open-source and available at https://github.com/SciYu/HiPhO.

Large language models (LLMs) face significant challenges with needle-in-a-haystack tasks, where relevant information ("the needle") must be drawn from a large pool of irrelevant context ("the haystack"). Previous studies have highlighted positional bias and distractor quantity as critical factors affecting model performance, yet the influence of gold context size has received little attention. We address this gap by systematically studying how variations in gold context length impact LLM performance on long-context question answering tasks. Our experiments reveal that LLM performance drops sharply when the gold context is shorter, i.e., smaller gold contexts consistently degrade model performance and amplify positional sensitivity, posing a major challenge for agentic systems that must integrate scattered, fine-grained information of varying lengths. This pattern holds across three diverse domains (general knowledge, biomedical reasoning, and mathematical reasoning) and seven state-of-the-art LLMs of various sizes and architectures. Our work provides clear insights to guide the design of robust, context-aware LLM-driven systems.

Large language models (LLMs) have grown in their usage to provide support for question answering across numerous disciplines. The models on their own have already shown promise for answering basic questions, however fail quickly where expert domain knowledge is required or the question is nuanced. Scientific research often involves searching for relevant literature, distilling pertinent information from that literature and analysing how the findings support or contradict one another. The information is often encapsulated in the full text body of research articles, rather than just in the abstracts. Statements within these articles frequently require the wider article context to be fully understood. We have built an LLM-based system that performs such search and distillation of information encapsulated in scientific literature, and we evaluate our keyword based search and information distillation system against a set of biology related questions from previously released literature benchmarks. We demonstrate sparse retrieval methods exhibit results close to state of the art without the need for dense retrieval, with its associated infrastructure and complexity overhead. We also show how to increase the coverage of relevant documents for literature review generation.

We introduce a large scale MAchine Reading COmprehension dataset, which we name MS MARCO. The dataset comprises of 1,010,916 anonymized questions---sampled from Bing's search query logs---each with a human generated answer and 182,669 completely human rewritten generated answers. In addition, the dataset contains 8,841,823 passages---extracted from 3,563,535 web documents retrieved by Bing---that provide the information necessary for curating the natural language answers. A question in the MS MARCO dataset may have multiple answers or no answers at all. Using this dataset, we propose three different tasks with varying levels of difficulty: (i) predict if a question is answerable given a set of context passages, and extract and synthesize the answer as a human would (ii) generate a well-formed answer (if possible) based on the context passages that can be understood with the question and passage context, and finally (iii) rank a set of retrieved passages given a question. The size of the dataset and the fact that the questions are derived from real user search queries distinguishes MS MARCO from other well-known publicly available datasets for machine reading comprehension and question-answering. We believe that the scale and the real-world nature of this dataset makes it attractive for benchmarking machine reading comprehension and question-answering models.

The VNHSGE (VietNamese High School Graduation Examination) dataset, developed exclusively for evaluating large language models (LLMs), is introduced in this article. The dataset, which covers nine subjects, was generated from the Vietnamese National High School Graduation Examination and comparable tests. 300 literary essays have been included, and there are over 19,000 multiple-choice questions on a range of topics. The dataset assesses LLMs in multitasking situations such as question answering, text generation, reading comprehension, visual question answering, and more by including both textual data and accompanying images. Using ChatGPT and BingChat, we evaluated LLMs on the VNHSGE dataset and contrasted their performance with that of Vietnamese students to see how well they performed. The results show that ChatGPT and BingChat both perform at a human level in a number of areas, including literature, English, history, geography, and civics education. They still have space to grow, though, especially in the areas of mathematics, physics, chemistry, and biology. The VNHSGE dataset seeks to provide an adequate benchmark for assessing the abilities of LLMs with its wide-ranging coverage and variety of activities. We intend to promote future developments in the creation of LLMs by making this dataset available to the scientific community, especially in resolving LLMs' limits in disciplines involving mathematics and the natural sciences.

Finding relevant literature underpins the practice of evidence-based medicine. From 2014 to 2016, TREC conducted a clinical decision support track, wherein participants were tasked with finding articles relevant to clinical questions posed by physicians. In total, 87 teams have participated over the past three years, generating 395 runs. During this period, each team has trialled a variety of methods. While there was significant overlap in the methods employed by different teams, the results were varied. Due to the diversity of the platforms used, the results arising from the different techniques are not directly comparable, reducing the ability to build on previous work. By using a stable platform, we have been able to compare different document and query processing techniques, allowing us to experiment with different search parameters. We have used our system to reproduce leading teams runs, and compare the results obtained. By benchmarking our indexing and search techniques, we can statistically test a variety of hypotheses, paving the way for further research.

Context utilisation, the ability of Language Models (LMs) to incorporate relevant information from the provided context when generating responses, remains largely opaque to users, who cannot determine whether models draw from parametric memory or provided context, nor identify which specific context pieces inform the response. Highlight explanations (HEs) offer a natural solution as they can point the exact context pieces and tokens that influenced model outputs. However, no existing work evaluates their effectiveness in accurately explaining context utilisation. We address this gap by introducing the first gold standard HE evaluation framework for context attribution, using controlled test cases with known ground-truth context usage, which avoids the limitations of existing indirect proxy evaluations. To demonstrate the framework's broad applicability, we evaluate four HE methods -- three established techniques and MechLight, a mechanistic interpretability approach we adapt for this task -- across four context scenarios, four datasets, and five LMs. Overall, we find that MechLight performs best across all context scenarios. However, all methods struggle with longer contexts and exhibit positional biases, pointing to fundamental challenges in explanation accuracy that require new approaches to deliver reliable context utilisation explanations at scale.

In the era of Large Language Models (LLMs), human-computer interaction has evolved towards natural language, offering unprecedented flexibility. Despite this, LLMs are heavily reliant on well-structured prompts to function efficiently within the realm of In-Context Learning. Vanilla In-Context Learning relies on human-provided contexts, such as labeled examples, explicit instructions, or other guiding mechanisms that shape the model's outputs. To address this challenge, our study presents a universal framework named Automatic In-Context Learning. Upon receiving a user's request, we ask the model to independently generate examples, including labels, instructions, or reasoning pathways. The model then leverages this self-produced context to tackle the given problem. Our approach is universally adaptable and can be implemented in any setting where vanilla In-Context Learning is applicable. We demonstrate that our method yields strong performance across a range of tasks, standing up well when compared to existing methods.

Best practices for high conflict conversations like counseling or customer support almost always include recommendations to paraphrase the previous speaker. Although paraphrase classification has received widespread attention in NLP, paraphrases are usually considered independent from context, and common models and datasets are not applicable to dialog settings. In this work, we investigate paraphrases in dialog (e.g., Speaker 1: "That book is mine." becomes Speaker 2: "That book is yours."). We provide an operationalization of context-dependent paraphrases, and develop a training for crowd-workers to classify paraphrases in dialog. We introduce a dataset with utterance pairs from NPR and CNN news interviews annotated for context-dependent paraphrases. To enable analyses on label variation, the dataset contains 5,581 annotations on 600 utterance pairs. We present promising results with in-context learning and with token classification models for automatic paraphrase detection in dialog.

The ability of language models to learn a task from a few examples in context has generated substantial interest. Here, we provide a perspective that situates this type of supervised few-shot learning within a much broader spectrum of meta-learned in-context learning. Indeed, we suggest that any distribution of sequences in which context non-trivially decreases loss on subsequent predictions can be interpreted as eliciting a kind of in-context learning. We suggest that this perspective helps to unify the broad set of in-context abilities that language models exhibit x2014 such as adapting to tasks from instructions or role play, or extrapolating time series. This perspective also sheds light on potential roots of in-context learning in lower-level processing of linguistic dependencies (e.g. coreference or parallel structures). Finally, taking this perspective highlights the importance of generalization, which we suggest can be studied along several dimensions: not only the ability to learn something novel, but also flexibility in learning from different presentations, and in applying what is learned. We discuss broader connections to past literature in meta-learning and goal-conditioned agents, and other perspectives on learning and adaptation. We close by suggesting that research on in-context learning should consider this broader spectrum of in-context capabilities and types of generalization.

Mathematical questioning is crucial for assessing students problem-solving skills. Since manually creating such questions requires substantial effort, automatic methods have been explored. Existing state-of-the-art models rely on fine-tuning strategies and struggle to generate questions that heavily involve multiple steps of logical and arithmetic reasoning. Meanwhile, large language models(LLMs) such as ChatGPT have excelled in many NLP tasks involving logical and arithmetic reasoning. Nonetheless, their applications in generating educational questions are underutilized, especially in the field of mathematics. To bridge this gap, we take the first step to conduct an in-depth analysis of ChatGPT in generating pre-university math questions. Our analysis is categorized into two main settings: context-aware and context-unaware. In the context-aware setting, we evaluate ChatGPT on existing math question-answering benchmarks covering elementary, secondary, and ternary classes. In the context-unaware setting, we evaluate ChatGPT in generating math questions for each lesson from pre-university math curriculums that we crawl. Our crawling results in TopicMath, a comprehensive and novel collection of pre-university math curriculums collected from 121 math topics and 428 lessons from elementary, secondary, and tertiary classes. Through this analysis, we aim to provide insight into the potential of ChatGPT as a math questioner.

With the increasing ability of large language models (LLMs), in-context learning (ICL) has become a new paradigm for natural language processing (NLP), where LLMs make predictions only based on contexts augmented with a few examples. It has been a new trend to explore ICL to evaluate and extrapolate the ability of LLMs. In this paper, we aim to survey and summarize the progress and challenges of ICL. We first present a formal definition of ICL and clarify its correlation to related studies. Then, we organize and discuss advanced techniques, including training strategies, demonstration designing strategies, as well as related analysis. Finally, we discuss the challenges of ICL and provide potential directions for further research. We hope that our work can encourage more research on uncovering how ICL works and improving ICL.

The role of large language models (LLMs) in education is an increasing area of interest today, considering the new opportunities they offer for teaching, learning, and assessment. This cutting-edge tutorial provides an overview of the educational applications of NLP and the impact that the recent advances in LLMs have had on this field. We will discuss the key challenges and opportunities presented by LLMs, grounding them in the context of four major educational applications: reading, writing, and speaking skills, and intelligent tutoring systems (ITS). This COLING 2025 tutorial is designed for researchers and practitioners interested in the educational applications of NLP and the role LLMs have to play in this area. It is the first of its kind to address this timely topic.

Software testing is a critical aspect of software development, yet generating test cases remains a routine task for engineers. This paper presents a benchmark, CLOVER, to evaluate models' capabilities in generating and completing test cases under specific conditions. Spanning from simple assertion completions to writing test cases that cover specific code blocks across multiple files, these tasks are based on 12 python repositories, analyzing 845 problems with context lengths ranging from 4k to 128k tokens. Utilizing code testing frameworks, we propose a method to construct retrieval contexts using coverage information. While models exhibit comparable performance with short contexts, notable differences emerge with 16k contexts. Notably, models like GPT-4o and Claude 3.5 can effectively leverage relevant snippets; however, all models score below 35\% on the complex Task III, even with the oracle context provided, underscoring the benchmark's significance and the potential for model improvement. The benchmark is containerized for code execution across tasks, and we will release the code, data, and construction methodologies.

A large number of reading comprehension (RC) datasets has been created recently, but little analysis has been done on whether they generalize to one another, and the extent to which existing datasets can be leveraged for improving performance on new ones. In this paper, we conduct such an investigation over ten RC datasets, training on one or more source RC datasets, and evaluating generalization, as well as transfer to a target RC dataset. We analyze the factors that contribute to generalization, and show that training on a source RC dataset and transferring to a target dataset substantially improves performance, even in the presence of powerful contextual representations from BERT (Devlin et al., 2019). We also find that training on multiple source RC datasets leads to robust generalization and transfer, and can reduce the cost of example collection for a new RC dataset. Following our analysis, we propose MultiQA, a BERT-based model, trained on multiple RC datasets, which leads to state-of-the-art performance on five RC datasets. We share our infrastructure for the benefit of the research community.

We identify two crucial limitations in the evaluation of recent parallel-integrated method Parallel Context Windows (PCW), which extends the maximum context lengths of language models, e.g., 2048 for LLaMA, by harnessing window-wise attention and positional embedding techniques. We first show that a simple yet strong baseline, weighted sum ensemble, is missing for the in-context few-shot classification. Moreover, on more challenging Chain-of-Thought (CoT) reasoning (e.g., HotpotQA), PCW would present unexpected deterioration regarding question miscomprehension and false inference. Based on our findings, we suggest that the existing PCW design may not guarantee sufficient improvement and practicality in handling lengthy documents in real-world applications. More community efforts on enabling language models' long context understanding ability should be paid.

Language model users often issue queries that lack specification, where the context under which a query was issued -- such as the user's identity, the query's intent, and the criteria for a response to be useful -- is not explicit. For instance, a good response to a subjective query like "What book should I read next?" would depend on the user's preferences, and a good response to an open-ended query like "How do antibiotics work against bacteria?" would depend on the user's expertise. This makes evaluation of responses to such queries an ill-posed task, as evaluators may make arbitrary judgments about the response quality. To remedy this, we present contextualized evaluations, a protocol that synthetically constructs context surrounding an underspecified query and provides it during evaluation. We find that the presence of context can 1) alter conclusions drawn from evaluation, even flipping win rates between model pairs, 2) nudge evaluators to make fewer judgments based on surface-level criteria, like style, and 3) provide new insights about model behavior across diverse contexts. Specifically, our procedure uncovers an implicit bias towards WEIRD contexts in models' "default" responses and we find that models are not equally sensitive to following different contexts, even when they are provided in prompts.

Broad textual understanding and in-context learning require language models that utilize full document contexts. Due to the implementation challenges associated with directly training long-context models, many methods have been proposed for extending models to handle long contexts. However, owing to differences in data and model classes, it has been challenging to compare these approaches, leading to uncertainty as to how to evaluate long-context performance and whether it differs from standard evaluation. We implement a controlled protocol for extension methods with a standardized evaluation, utilizing consistent base models and extension data. Our study yields several insights into long-context behavior. First, we reaffirm the critical role of perplexity as a general-purpose performance indicator even in longer-context tasks. Second, we find that current approximate attention methods systematically underperform across long-context tasks. Finally, we confirm that exact fine-tuning based methods are generally effective within the range of their extension, whereas extrapolation remains challenging. All codebases, models, and checkpoints will be made available open-source, promoting transparency and facilitating further research in this critical area of AI development.

Describes an audio dataset of spoken words designed to help train and evaluate keyword spotting systems. Discusses why this task is an interesting challenge, and why it requires a specialized dataset that is different from conventional datasets used for automatic speech recognition of full sentences. Suggests a methodology for reproducible and comparable accuracy metrics for this task. Describes how the data was collected and verified, what it contains, previous versions and properties. Concludes by reporting baseline results of models trained on this dataset.

We introduce WIQA, the first large-scale dataset of "What if..." questions over procedural text. WIQA contains three parts: a collection of paragraphs each describing a process, e.g., beach erosion; a set of crowdsourced influence graphs for each paragraph, describing how one change affects another; and a large (40k) collection of "What if...?" multiple-choice questions derived from the graphs. For example, given a paragraph about beach erosion, would stormy weather result in more or less erosion (or have no effect)? The task is to answer the questions, given their associated paragraph. WIQA contains three kinds of questions: perturbations to steps mentioned in the paragraph; external (out-of-paragraph) perturbations requiring commonsense knowledge; and irrelevant (no effect) perturbations. We find that state-of-the-art models achieve 73.8% accuracy, well below the human performance of 96.3%. We analyze the challenges, in particular tracking chains of influences, and present the dataset as an open challenge to the community.

In-Context Learning (ICL) is a technique by which language models make predictions based on examples provided in their input context. Previously, their context window size imposed a limit on the number of examples that can be shown, making example selection techniques crucial for identifying the maximally effective set of examples. However, the recent advent of Long Context Language Models (LCLMs) has significantly increased the number of examples that can be included in context, raising an important question of whether ICL performance in a many-shot regime is still sensitive to the method of sample selection. To answer this, we revisit these approaches in the context of LCLMs through extensive experiments on 18 datasets spanning 4 tasks. Surprisingly, we observe that sophisticated example selection techniques do not yield significant improvements over a simple random sample selection method. Instead, we find that the advent of LCLMs has fundamentally shifted the challenge of ICL from that of selecting the most effective examples to that of collecting sufficient examples to fill the context window. Specifically, in certain datasets, including all available examples does not fully utilize the context window; however, by augmenting the examples in context with a simple data augmentation approach, we substantially improve ICL performance by 5%.

We extend the context length of Llama-3-8B-Instruct from 8K to 80K via QLoRA fine-tuning. The entire training cycle is super efficient, which takes 8 hours on one 8xA800 (80G) GPU machine. The resulted model exhibits superior performances across a broad range of evaluation tasks, such as NIHS, topic retrieval, and long-context language understanding; meanwhile, it also well preserves the original capability over short contexts. The dramatic context extension is mainly attributed to merely 3.5K synthetic training samples generated by GPT-4 , which indicates the LLMs' inherent (yet largely underestimated) potential to extend its original context length. In fact, the context length could be extended far beyond 80K with more computation resources. Therefore, the team will publicly release the entire resources (including data, model, data generation pipeline, training code) so as to facilitate the future research from the community: https://github.com/FlagOpen/FlagEmbedding.

In-Context Learning (ICL) enables transformer-based language models to adapt to new tasks by conditioning on demonstration examples. However, traditional example-driven in-context learning lacks explicit modules for knowledge retrieval and transfer at the abstraction level. Inspired by cognitive science, specifically schema theory, which holds that humans interpret new information by activating pre-existing mental frameworks (schemas) to structure understanding, we introduce SCHEMA ACTIVATED IN CONTEXT LEARNING (SA-ICL). This framework extracts the representation of the building blocks of cognition for the reasoning process instilled from prior examples, creating an abstracted schema, a lightweight, structured template of key inferential steps and their relationships, which is then used to augment a model's reasoning process when presented with a novel question. We demonstrate that a broad range of large language models (LLMs) lack the capacity to form and utilize internal schema-based learning representations implicitly, but instead benefit significantly from explicit schema-based scaffolding. Across chemistry and physics questions from the GPQA dataset, our experiments show that SA-ICL consistently boosts performance, up to 36.19 percent, when the single demonstration example is of high quality, which simultaneously reduces reliance on the number of demonstrations and enhances interpretability. SCHEMA ACTIVATED IN CONTEXT LEARNING not only bridges disparate ICL strategies ranging from pattern priming to Chain-of-Thought prompting, but also paves a new path for enhancing human-like reasoning in LLMs.

Assessing the effectiveness of large language models (LLMs) in addressing diverse tasks is essential for comprehending their strengths and weaknesses. Conventional evaluation techniques typically apply a single prompting strategy uniformly across datasets, not considering the varying degrees of task complexity. We introduce the Hierarchical Prompting Taxonomy (HPT), a taxonomy that employs a Hierarchical Prompt Framework (HPF) composed of five unique prompting strategies, arranged from the simplest to the most complex, to assess LLMs more precisely and to offer a clearer perspective. This taxonomy assigns a score, called the Hierarchical Prompting Score (HP-Score), to datasets as well as LLMs based on the rules of the taxonomy, providing a nuanced understanding of their ability to solve diverse tasks and offering a universal measure of task complexity. Additionally, we introduce the Adaptive Hierarchical Prompt framework, which automates the selection of appropriate prompting strategies for each task. This study compares manual and adaptive hierarchical prompt frameworks using four instruction-tuned LLMs, namely Llama 3 8B, Phi 3 3.8B, Mistral 7B, and Gemma 7B, across four datasets: BoolQ, CommonSenseQA (CSQA), IWSLT-2017 en-fr (IWSLT), and SamSum. Experiments demonstrate the effectiveness of HPT, providing a reliable way to compare different tasks and LLM capabilities. This paper leads to the development of a universal evaluation metric that can be used to evaluate both the complexity of the datasets and the capabilities of LLMs. The implementation of both manual HPF and adaptive HPF is publicly available.

With the help of in-context learning (ICL), large language models (LLMs) have achieved impressive performance across various tasks. However, the function of descriptive instructions during ICL remains under-explored. In this work, we propose an ensemble prompt framework to describe the selection criteria of multiple in-context examples, and preliminary experiments on machine translation (MT) across six translation directions confirm that this framework boosts ICL perfromance. But to our surprise, LLMs might not necessarily care what the descriptions actually say, and the performance gain is primarily caused by the ensemble format, since the framework could lead to improvement even with random descriptive nouns. We further apply this new ensemble prompt on a range of commonsense, math, logical reasoning and hallucination tasks with three LLMs and achieve promising results, suggesting again that designing a proper prompt format would be much more effective and efficient than paying effort into specific descriptions. Our code will be publicly available once this paper is published.

Characterizing a large language model's (LLM's) knowledge of a given question is challenging. As a result, prior work has primarily examined LLM behavior under knowledge conflicts, where the model's internal parametric memory contradicts information in the external context. However, this does not fully reflect how well the model knows the answer to the question. In this paper, we first introduce a taxonomy of five knowledge statuses based on the consistency and correctness of LLM knowledge modes. We then propose KScope, a hierarchical framework of statistical tests that progressively refines hypotheses about knowledge modes and characterizes LLM knowledge into one of these five statuses. We apply KScope to nine LLMs across four datasets and systematically establish: (1) Supporting context narrows knowledge gaps across models. (2) Context features related to difficulty, relevance, and familiarity drive successful knowledge updates. (3) LLMs exhibit similar feature preferences when partially correct or conflicted, but diverge sharply when consistently wrong. (4) Context summarization constrained by our feature analysis, together with enhanced credibility, further improves update effectiveness and generalizes across LLMs.

Code completion aims at speeding up code writing by recommending to developers the next tokens they are likely to type. Deep Learning (DL) models pushed the boundaries of code completion by redefining what these coding assistants can do: We moved from predicting few code tokens to automatically generating entire functions. One important factor impacting the performance of DL-based code completion techniques is the context provided as input. With "context" we refer to what the model knows about the code to complete. In a simple scenario, the DL model might be fed with a partially implemented function to complete. In this case, the context is represented by the incomplete function and, based on it, the model must generate a prediction. It is however possible to expand such a context to include additional information, like the whole source code file containing the function to complete, which could be useful to boost the prediction performance. In this work, we present an empirical study investigating how the performance of a DL-based code completion technique is affected by different contexts. We experiment with 8 types of contexts and their combinations. These contexts include: (i) coding contexts, featuring information extracted from the code base in which the code completion is invoked (e.g., code components structurally related to the one to "complete"); (ii) process context, with information aimed at depicting the current status of the project in which a code completion task is triggered (e.g., a textual representation of open issues relevant for the code to complete); and (iii) developer contexts, capturing information about the developer invoking the code completion (e.g., the APIs frequently used). Our results show that additional contextual information can benefit the performance of DL-based code completion, with relative improvements up to +22% in terms of correct predictions.

The behavior and decision making of groups or communities can be dramatically influenced by individuals pushing particular agendas, e.g., to promote or disparage a person or an activity, to call for action, etc.. In the examination of online influence campaigns, particularly those related to important political and social events, scholars often concentrate on identifying the sources responsible for setting and controlling the agenda (e.g., public media). In this article we present a methodology for detecting specific instances of agenda control through social media where annotated data is limited or non-existent. By using a modest corpus of Twitter messages centered on the 2022 French Presidential Elections, we carry out a comprehensive evaluation of various approaches and techniques that can be applied to this problem. Our findings demonstrate that by treating the task as a textual entailment problem, it is possible to overcome the requirement for a large annotated training dataset.

This article details the advances made to a system that uses artificial intelligence to identify alarming student responses. This system is built into our assessment platform to assess whether a student's response indicates they are a threat to themselves or others. Such responses may include details concerning threats of violence, severe depression, suicide risks, and descriptions of abuse. Driven by advances in natural language processing, the latest model is a fine-tuned language model trained on a large corpus consisting of student responses and supplementary texts. We demonstrate that the use of a language model delivers a substantial improvement in accuracy over the previous iterations of this system.

We present RACE, a new dataset for benchmark evaluation of methods in the reading comprehension task. Collected from the English exams for middle and high school Chinese students in the age range between 12 to 18, RACE consists of near 28,000 passages and near 100,000 questions generated by human experts (English instructors), and covers a variety of topics which are carefully designed for evaluating the students' ability in understanding and reasoning. In particular, the proportion of questions that requires reasoning is much larger in RACE than that in other benchmark datasets for reading comprehension, and there is a significant gap between the performance of the state-of-the-art models (43%) and the ceiling human performance (95%). We hope this new dataset can serve as a valuable resource for research and evaluation in machine comprehension. The dataset is freely available at http://www.cs.cmu.edu/~glai1/data/race/ and the code is available at https://github.com/qizhex/RACE_AR_baselines.

Current long-context benchmarks primarily focus on retrieval-based tests, requiring Large Language Models (LLMs) to locate specific information within extensive input contexts, such as the needle-in-a-haystack (NIAH) benchmark. Long-context generation refers to the ability of a language model to generate coherent and contextually accurate text that spans across lengthy passages or documents. While recent studies show strong performance on NIAH and other retrieval-based long-context benchmarks, there is a significant lack of benchmarks for evaluating long-context generation capabilities. To bridge this gap and offer a comprehensive assessment, we introduce a synthetic benchmark, LongGenBench, which allows for flexible configurations of customized generation context lengths. LongGenBench advances beyond traditional benchmarks by redesigning the format of questions and necessitating that LLMs respond with a single, cohesive long-context answer. Upon extensive evaluation using LongGenBench, we observe that: (1) both API accessed and open source models exhibit performance degradation in long-context generation scenarios, ranging from 1.2% to 47.1%; (2) different series of LLMs exhibit varying trends of performance degradation, with the Gemini-1.5-Flash model showing the least degradation among API accessed models, and the Qwen2 series exhibiting the least degradation in LongGenBench among open source models.

Since the launch of ChatGPT in late 2022, the capacities of Large Language Models and their evaluation have been in constant discussion and evaluation both in academic research and in the industry. Scenarios and benchmarks have been developed in several areas such as law, medicine and math (Bommasani et al., 2023) and there is continuous evaluation of model variants. One area that has not received sufficient scenario development attention is journalism, and in particular journalistic sourcing and ethics. Journalism is a crucial truth-determination function in democracy (Vincent, 2023), and sourcing is a crucial pillar to all original journalistic output. Evaluating the capacities of LLMs to annotate stories for the different signals of sourcing and how reporters justify them is a crucial scenario that warrants a benchmark approach. It offers potential to build automated systems to contrast more transparent and ethically rigorous forms of journalism with everyday fare. In this paper we lay out a scenario to evaluate LLM performance on identifying and annotating sourcing in news stories on a five-category schema inspired from journalism studies (Gans, 2004). We offer the use case, our dataset and metrics and as the first step towards systematic benchmarking. Our accuracy findings indicate LLM-based approaches have more catching to do in identifying all the sourced statements in a story, and equally, in matching the type of sources. An even harder task is spotting source justifications.

Recent large language models (LLMs) have shown indications of mathematical reasoning ability. However it has not been clear how they would fare on more challenging competition-level problems. And while self-generated verbalizations of intermediate reasoning steps (i.e., chain-of-thought prompting) have been shown to be helpful, whether LLMs can make use of helpful side information such as problem-specific hints has not been investigated before. In this paper, we propose a challenging benchmark dataset for enabling such analyses. The Concept and Hint-Annotated Math Problems (CHAMP) consists of high school math competition problems, annotated with concepts, or general math facts, and hints, or problem-specific tricks. These annotations allow us to explore the effects of additional information, such as relevant hints, misleading concepts, or related problems. This benchmark is difficult, with the best model only scoring 58.1% in standard settings. With concepts and hints, performance sometimes improves, indicating that some models can make use of such side information. We further annotate model-generated solutions for their correctness. Using this corpus, we find that models often arrive at the correct final answer through wrong reasoning steps. In addition, we test whether models are able to verify these solutions, and find that most models struggle. The dataset and code are available on the project website.

As large language models continue to advance, their application in educational contexts remains underexplored and under-optimized. In this paper, we address this gap by introducing the first diverse benchmark tailored for educational scenarios, incorporating synthetic data containing 9 major scenarios and over 4,000 distinct educational contexts. To enable comprehensive assessment, we propose a set of multi-dimensional evaluation metrics that cover 12 critical aspects relevant to both teachers and students. We further apply human annotation to ensure the effectiveness of the model-generated evaluation responses. Additionally, we succeed to train a relatively small-scale model on our constructed dataset and demonstrate that it can achieve performance comparable to state-of-the-art large models (e.g., Deepseek V3, Qwen Max) on the test set. Overall, this work provides a practical foundation for the development and evaluation of education-oriented language models. Code and data are released at https://github.com/ybai-nlp/EduBench.

This paper presents teaching materials, particularly assignments and ideas for classroom activities, from a new course on large language models (LLMs) taught at Charles University. The assignments include experiments with LLM inference for weather report generation and machine translation. The classroom activities include class quizzes, focused research on downstream tasks and datasets, and an interactive "best paper" session aimed at reading and comprehension of research papers.

This paper explores the elusive mechanism underpinning in-context learning in Large Language Models (LLMs). Our work provides a novel perspective by examining in-context learning via the lens of surface repetitions. We quantitatively investigate the role of surface features in text generation, and empirically establish the existence of token co-occurrence reinforcement, a principle that strengthens the relationship between two tokens based on their contextual co-occurrences. By investigating the dual impacts of these features, our research illuminates the internal workings of in-context learning and expounds on the reasons for its failures. This paper provides an essential contribution to the understanding of in-context learning and its potential limitations, providing a fresh perspective on this exciting capability.

Recent advances in long-context large language models (LLMs) have led to the emerging paradigm of many-shot in-context learning (ICL), where it is observed that scaling many more demonstrating examples beyond the conventional few-shot setup in the context can lead to performance benefits. However, despite its promise, it is unclear what aspects dominate the benefits and whether simply scaling to more examples is the most effective way of improving many-shot ICL. In this work, we first provide an analysis of the factors driving many-shot ICL, and we find that 1) many-shot performance can still be attributed to often a few disproportionately influential examples and 2) identifying such influential examples ("optimize") and using them as demonstrations to regenerate new examples ("generate") can lead to further improvements. Inspired by the findings, we propose BRIDGE, an algorithm that alternates between the optimize step with Bayesian optimization to discover the influential sets of examples and the generate step to reuse this set to expand the reasoning paths of the examples back to the many-shot regime automatically. On Gemini, Claude, and Mistral LLMs of different sizes, we show that BRIDGE to significant improvements across a diverse set of tasks, including symbolic reasoning, numerical reasoning, and code generation.

Mathematical reasoning in real-world video settings presents a fundamentally different challenge than in static images or text. It requires interpreting fine-grained visual information, accurately reading handwritten or digital text, and integrating spoken cues, often dispersed non-linearly over time. In such multimodal contexts, success hinges not just on perception, but on selectively identifying and integrating the right contextual details from a rich and noisy stream of content. To this end, we introduce VideoMathQA, a benchmark designed to evaluate whether models can perform such temporally extended cross-modal reasoning on videos. The benchmark spans 10 diverse mathematical domains, covering videos ranging from 10 seconds to over 1 hour. It requires models to interpret structured visual content, understand instructional narratives, and jointly ground concepts across visual, audio, and textual modalities. We employ graduate-level experts to ensure high quality, totaling over 920 man-hours of annotation. To reflect real-world scenarios, questions are designed around three core reasoning challenges: direct problem solving, where answers are grounded in the presented question; conceptual transfer, which requires applying learned methods to new problems; and deep instructional comprehension, involving multi-step reasoning over extended explanations and partially worked-out solutions. Each question includes multi-step reasoning annotations, enabling fine-grained diagnosis of model capabilities. Through this benchmark, we highlight the limitations of existing approaches and establish a systematic evaluation framework for models that must reason, rather than merely perceive, across temporally extended and modality-rich mathematical problem settings. Our benchmark and evaluation code are available at: https://mbzuai-oryx.github.io/VideoMathQA

It is well documented that NLP models learn social biases, but little work has been done on how these biases manifest in model outputs for applied tasks like question answering (QA). We introduce the Bias Benchmark for QA (BBQ), a dataset of question sets constructed by the authors that highlight attested social biases against people belonging to protected classes along nine social dimensions relevant for U.S. English-speaking contexts. Our task evaluates model responses at two levels: (i) given an under-informative context, we test how strongly responses reflect social biases, and (ii) given an adequately informative context, we test whether the model's biases override a correct answer choice. We find that models often rely on stereotypes when the context is under-informative, meaning the model's outputs consistently reproduce harmful biases in this setting. Though models are more accurate when the context provides an informative answer, they still rely on stereotypes and average up to 3.4 percentage points higher accuracy when the correct answer aligns with a social bias than when it conflicts, with this difference widening to over 5 points on examples targeting gender for most models tested.

Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty. Such "hallucinations" persist even in state-of-the-art systems and undermine trust. We argue that language models hallucinate because the training and evaluation procedures reward guessing over acknowledging uncertainty, and we analyze the statistical causes of hallucinations in the modern training pipeline. Hallucinations need not be mysterious -- they originate simply as errors in binary classification. If incorrect statements cannot be distinguished from facts, then hallucinations in pretrained language models will arise through natural statistical pressures. We then argue that hallucinations persist due to the way most evaluations are graded -- language models are optimized to be good test-takers, and guessing when uncertain improves test performance. This "epidemic" of penalizing uncertain responses can only be addressed through a socio-technical mitigation: modifying the scoring of existing benchmarks that are misaligned but dominate leaderboards, rather than introducing additional hallucination evaluations. This change may steer the field toward more trustworthy AI systems.

We explore the use of long-context capabilities in large language models to create synthetic reading comprehension data from entire books. Previous efforts to construct such datasets relied on crowd-sourcing, but the emergence of transformers with a context size of 1 million or more tokens now enables entirely automatic approaches. Our objective is to test the capabilities of LLMs to analyze, understand, and reason over problems that require a detailed comprehension of long spans of text, such as questions involving character arcs, broader themes, or the consequences of early actions later in the story. We propose a holistic pipeline for automatic data generation including question generation, answering, and model scoring using an ``Evaluator''. We find that a relative approach, comparing answers between models in a pairwise fashion and ranking with a Bradley-Terry model, provides a more consistent and differentiating scoring mechanism than an absolute scorer that rates answers individually. We also show that LLMs from different model families produce moderate agreement in their ratings. We ground our approach using the manually curated NarrativeQA dataset, where our evaluator shows excellent agreement with human judgement and even finds errors in the dataset. Using our automatic evaluation approach, we show that using an entire book as context produces superior reading comprehension performance compared to baseline no-context (parametric knowledge only) and retrieval-based approaches.

Large language models are prominently used in real-world applications, often tasked with reasoning over large volumes of documents. An exciting development in this space is models boasting extended context capabilities, with some accommodating over 2 million tokens. Such long context model capabilities remain uncertain in production systems, motivating the need to benchmark their performance on real world use cases. We address this challenge by proposing SWiM, an evaluation framework that addresses the limitations of standard tests. Testing the framework on eight long context models, we find that even strong models such as GPT-4 and Claude 3 Opus degrade in performance when information is present in the middle of the context window (lost-in-the-middle effect). Next, in addition to our benchmark, we propose medoid voting, a simple, but effective training-free approach that helps alleviate this effect, by generating responses a few times, each time randomly permuting documents in the context, and selecting the medoid answer. We evaluate medoid voting on single document QA tasks, achieving up to a 24% lift in accuracy.

Large language models have made significant advancements in natural language processing (NLP), exhibiting human performance across various classic NLP tasks. These tasks, however, focus on structure and semantics, and few are designed to assess reasoning abilities and real-world knowledge, which are increasingly vital given that these models are trained on extensive textual data and information. While prior research primarily focuses on English, in this work, we gather a collection of exam problems from primary school to university entrance tests in Indonesia, and evaluate whether large language models can pass the exams. We obtain 14,906 questions across 63 tasks and levels, with 46\% of the questions focusing on assessing proficiency in the Indonesian language and knowledge of nine local languages and cultures in Indonesia. Our empirical evaluations show that GPT-3.5 only manages to pass the Indonesian primary school level, with limited knowledge of the Indonesian local languages and cultures. Other smaller models such as BLOOMZ and Falcon fail the exams.

Establishing whether language models can use contextual information in a human-plausible way is important to ensure their safe adoption in real-world settings. However, the questions of when and which parts of the context affect model generations are typically tackled separately, and current plausibility evaluations are practically limited to a handful of artificial benchmarks. To address this, we introduce Plausibility Evaluation of Context Reliance (PECoRe), an end-to-end interpretability framework designed to quantify context usage in language models' generations. Our approach leverages model internals to (i) contrastively identify context-sensitive target tokens in generated texts and (ii) link them to contextual cues justifying their prediction. We use PECoRe to quantify the plausibility of context-aware machine translation models, comparing model rationales with human annotations across several discourse-level phenomena. Finally, we apply our method to unannotated generations to identify context-mediated predictions and highlight instances of (im)plausible context usage in model translations.

The long-context capabilities of large language models (LLMs) have been a hot topic in recent years. To evaluate the performance of LLMs in different scenarios, various assessment benchmarks have emerged. However, as most of these benchmarks focus on identifying key information to answer questions, which mainly requires the retrieval ability of LLMs, these benchmarks can partially represent the reasoning performance of LLMs from large amounts of information. Meanwhile, although LLMs often claim to have context windows of 32k, 128k, 200k, or even longer, these benchmarks fail to reveal the actual supported length of these LLMs. To address these issues, we propose the LongIns benchmark dataset, a challenging long-context instruction-based exam for LLMs, which is built based on the existing instruction datasets. Specifically, in our LongIns, we introduce three evaluation settings: Global Instruction & Single Task (GIST), Local Instruction & Single Task (LIST), and Local Instruction & Multiple Tasks (LIMT). Based on LongIns, we perform comprehensive evaluations on existing LLMs and have the following important findings: (1). The top-performing GPT-4 with 128k context length performs poorly on the evaluation context window of 16k in our LongIns. (2). For the multi-hop reasoning ability of many existing LLMs, significant efforts are still needed under short context windows (less than 4k).

We present the Stanford Question Answering Dataset (SQuAD), a new reading comprehension dataset consisting of 100,000+ questions posed by crowdworkers on a set of Wikipedia articles, where the answer to each question is a segment of text from the corresponding reading passage. We analyze the dataset to understand the types of reasoning required to answer the questions, leaning heavily on dependency and constituency trees. We build a strong logistic regression model, which achieves an F1 score of 51.0%, a significant improvement over a simple baseline (20%). However, human performance (86.8%) is much higher, indicating that the dataset presents a good challenge problem for future research. The dataset is freely available at https://stanford-qa.com