Daily Papers

Test data from public benchmarks inevitably leaks into pretraining corpora, inflating evaluation scores once memorized. Contamination mitigation evaluation intervenes in the decoding process to suppress memorization and restore a contaminated model's genuine capability, but its prevailing metric, the G-AP (Gap of Aggregate Performance), is flawed. Discrete correct/incorrect readouts cannot characterize per-question performance, averaging before differencing lets over- and under-suppression cancel out, and uniform per-question weighting invites strategies to push solve probabilities onto the clean model's high-frequency values. We propose SA-PPG (Stratified Aggregate of Per-question Probability Gaps): estimate each question's solve probability by sampling, difference it against the clean model per question, and aggregate within groups defined by the clean model's solve probability. Existing mitigation strategies first estimate where contamination lies and then operate on the estimate, so they are only as correct as the estimate. RailCap instead judges contamination during generation: whenever a sample falls back onto the greedy trajectory, the next trajectory token is capped to the runner-up, accumulating suppression until the response distribution becomes sufficiently dispersed. Across multiple contaminated models and benchmarks, SA-PPG reveals that prior strategies' restoration is substantially overestimated, while RailCap attains the lowest SA-PPG.

We investigate the ability of language models to perform compositional reasoning tasks where the overall solution depends on correctly composing the answers to sub-problems. We measure how often models can correctly answer all sub-problems but not generate the overall solution, a ratio we call the compositionality gap. We evaluate this ratio by asking multi-hop questions with answers that require composing multiple facts unlikely to have been observed together during pretraining. In the GPT-3 family of models, as model size increases we show that the single-hop question answering performance improves faster than the multi-hop performance does, therefore the compositionality gap does not decrease. This surprising result suggests that while more powerful models memorize and recall more factual knowledge, they show no corresponding improvement in their ability to perform this kind of compositional reasoning. We then demonstrate how elicitive prompting (such as chain of thought) narrows the compositionality gap by reasoning explicitly instead of implicitly. We present a new method, self-ask, that further improves on chain of thought. In our method, the model explicitly asks itself (and then answers) follow-up questions before answering the initial question. We finally show that self-ask's structured prompting lets us easily plug in a search engine to answer the follow-up questions, which additionally improves accuracy.

In this paper, we propose the task of consecutive question generation (CQG), which generates a set of logically related question-answer pairs to understand a whole passage, with a comprehensive consideration of the aspects including accuracy, coverage, and informativeness. To achieve this, we first examine the four key elements of CQG, i.e., question, answer, rationale, and context history, and propose a novel dynamic multitask framework with one main task generating a question-answer pair, and four auxiliary tasks generating other elements. It directly helps the model generate good questions through both joint training and self-reranking. At the same time, to fully explore the worth-asking information in a given passage, we make use of the reranking losses to sample the rationales and search for the best question series globally. Finally, we measure our strategy by QA data augmentation and manual evaluation, as well as a novel application of generated question-answer pairs on DocNLI. We prove that our strategy can improve question generation significantly and benefit multiple related NLP tasks.

The handling of probabilities in the form of uncertainty or partial information is an essential task for LLMs in many settings and applications. A common approach to evaluate an LLM's probabilistic reasoning capabilities is to assess its ability to answer questions pertaining to probability through the use of multiple-choice questions (MCQs). However, this paradigm, which we refer to as explicit probabilistic reasoning, has been shown in the literature to yield significant limitations (e.g., sensitivity to answer ordering). In this work, we introduce an alternative approach, named implicit probabilistic reasoning, which evaluates the models' ability to integrate probabilistic reasoning into their text generation process. To achieve this, we rephrase MCQs as text-completion scenarios with a determined set of outcomes and compare the model's next-token probability assignments to the true likelihood of the outcomes. In line with previous work, we find that models exhibit solid performance in their explicit probabilistic reasoning (i.e., answers to MCQs). However, during text completion (i.e., implicit probabilistic reasoning), where the same information must be taken into account to generate text, the models' predictions often significantly diverge from the known ground truth. For instance, our evaluation method reveals that implicit probabilistic reasoning is improperly influenced by many factors, such as independent prior events, partial observations about a result, or statistical background information. All of these issues can cause erroneous results to be produced in text generation, which are not detected by conventional MCQ-based evaluation.

While Retrieval-Augmented Generation (RAG) mitigates hallucination and knowledge staleness in Large Language Models (LLMs), existing frameworks often falter on complex, multi-hop queries that require synthesizing information from disparate sources. Current advanced RAG methods, employing iterative or adaptive strategies, lack a robust mechanism to systematically identify and fill evidence gaps, often propagating noise or failing to gather a comprehensive context. We introduce FAIR-RAG, a novel agentic framework that transforms the standard RAG pipeline into a dynamic, evidence-driven reasoning process. At its core is an Iterative Refinement Cycle governed by a module we term Structured Evidence Assessment (SEA). The SEA acts as an analytical gating mechanism: it deconstructs the initial query into a checklist of required findings and audits the aggregated evidence to identify confirmed facts and, critically, explicit informational gaps. These gaps provide a precise signal to an Adaptive Query Refinement agent, which generates new, targeted sub-queries to retrieve missing information. This cycle repeats until the evidence is verified as sufficient, ensuring a comprehensive context for a final, strictly faithful generation. We conducted experiments on challenging multi-hop QA benchmarks, including HotpotQA, 2WikiMultiHopQA, and MusiQue. In a unified experimental setup, FAIR-RAG significantly outperforms strong baselines. On HotpotQA, it achieves an F1-score of 0.453 -- an absolute improvement of 8.3 points over the strongest iterative baseline -- establishing a new state-of-the-art for this class of methods on these benchmarks. Our work demonstrates that a structured, evidence-driven refinement process with explicit gap analysis is crucial for unlocking reliable and accurate reasoning in advanced RAG systems for complex, knowledge-intensive tasks.

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.

Knowledge graphs have been proposed as a structured alternative to flat retrieval-augmented generation for long-term agent memory, on the assumption that representing conversations as entities and relations improves recall. We evaluate that assumption directly. Our framework extracts each conversational turn into typed nodes and attributed edges, answers questions from a two-hop subgraph, and periodically prunes nodes that score low on a weighted combination of recency, access frequency, degree centrality, and age. On LongMemEval, the graph does not outperform a flat vector baseline at a matched candidate-generation budget of five retrieval roots: token F1 is 0.417 against 0.468, and a paired bootstrap over 500 questions gives Δ= -0.050 (95\% CI [-0.085, -0.016]). The gap is widest on questions that require recalling a specific prior assistant turn, where judged correctness falls from 0.911 to 0.607, suggesting that decomposing a turn into entities discards the surface form these questions depend on. The forgetting module is more successful. Applied once to a persistent 27{,}021-node graph, it removes 9.8\% of nodes and 9.5\% of stored bytes; token F1 is unchanged (+0.001, 95\% CI [-0.015, +0.016]) and judged correctness falls by 1.6 points, with the 95\% interval bounding any loss at 3.8 points ([-0.038, +0.006]). Because our extractor is a single small model evaluated on one benchmark, these results characterise this extraction-based pipeline rather than graph-structured memory in general. Code: https://github.com/skhanzad/Selective-Amnesia

Verification for retrieval-augmented generation usually scores each retrieved chunk and drops the ones that fail. We show this cannot work for multi-hop questions, and show what does. Per-chunk scoring assumes one chunk is a sufficient premise for the answer. Multi-hop questions are built so that none is, and the paragraph carrying the answer is the one the question does not name. Entailment scoring reaches 0.643, 0.523 and 0.560 AUC on HotpotQA, 2WikiMultihopQA and MuSiQue, against 0.951 on single-hop SQuAD. Seven controls rule out model capacity, premise length, hypothesis template, decision threshold, retriever, answer-matching criterion and prompt. End to end across three datasets, three generator sizes and two prompts, per-chunk gating is significantly worse than not filtering at all in every cell, and its penalty grows with generator capability. The repair is to condition verification on the decomposed sub-question rather than the original query. Using MuSiQue's gold decomposition, entailment on a later hop rises from 0.546, which is chance, to 0.840, a paired lift of +0.355 with a bootstrap interval of [0.331, 0.382]. An off-the-shelf Qwen2.5-7B decomposer, given the question and the top retrieved paragraph, reaches 0.637 and captures 31% of that ceiling; decomposing without retrieval reaches 0.533, below the original question. Iterative retrieval systems already produce such decompositions and discard them before verifying.

Recent work finds that retrieval-augmented generation with large language models is prone to be influenced by the order of retrieved documents in the context. However, the lack of in-depth analysis limits the use of this phenomenon for prompt engineering in practice. In this study, we posit that likelihoods serve as an effective gauge for language model performance. Through experiments on two question-answering datasets with a variety of state-of-the-art language models, we reveal correlations between answer accuracy and the likelihood of the question at both the corpus level and the instance level. In addition, we find that question likelihood can also indicate the position of the task-relevant information in the context. Based on these findings, we propose two methods that use question likelihood as a gauge for selecting and constructing prompts that lead to better performance. We demonstrate their effectiveness with experiments. In addition, our likelihood-based methods are efficient, as they only need to compute the likelihood of the input, requiring much fewer language model passes than heuristic prompt engineering methods that require generating responses. Our analysis deepens our understanding of how input prompts affect model performance and provides a promising direction for efficient prompt optimization.

We present NewsQA, a challenging machine comprehension dataset of over 100,000 human-generated question-answer pairs. Crowdworkers supply questions and answers based on a set of over 10,000 news articles from CNN, with answers consisting of spans of text from the corresponding articles. We collect this dataset through a four-stage process designed to solicit exploratory questions that require reasoning. A thorough analysis confirms that NewsQA demands abilities beyond simple word matching and recognizing textual entailment. We measure human performance on the dataset and compare it to several strong neural models. The performance gap between humans and machines (0.198 in F1) indicates that significant progress can be made on NewsQA through future research. The dataset is freely available at https://datasets.maluuba.com/NewsQA.

People overthink; language models over-sample, and the extra effort can talk both into a worse answer. Reasoning systems answer a hard question by sampling it many times (test-time scaling), and the more they draw, the more often a correct answer turns up somewhere, so coverage, the fraction of problems with at least one correct try, climbs and appears to be progress. But a deployed system must return one answer, and choosing it, not knowing which try is right, is selection; selection is capped, and past a point extra samples only make the model surer of a confident mistake, even as every draw adds cost. The gap between climbing coverage and stalled selection, the identifiability gap, is the answer a model can produce but not pick. So the real question is not whether to sample but how far, and the answer is: not far. For picking an answer, the vote has already settled within a few dozen draws, the modal ceiling; for scoring a benchmark, sooner still, the correlation ceiling. Beyond that, extra draws cost compute and add nothing, and can even make the answer worse. This paper turns the cutoff into a single number, the effective number of samples, that any sampling run already reveals. The bottleneck is recognizing a right answer, not generating one.

The zero-shot chain of thought (CoT) approach is often used in question answering (QA) by language models (LMs) for tasks that require multiple reasoning steps, typically enhanced by the prompt "Let's think step by step." However, some QA tasks hinge more on accessing relevant knowledge than on chaining reasoning steps. We introduce a simple general prompting technique, called PREP, that involves using two instances of LMs: the first (LM1) generates relevant information, and the second (LM2) answers the question based on this information. PREP is designed to be general and independent of the user's domain knowledge, making it applicable across various QA tasks without the need for specialized prompt engineering. To evaluate the effectiveness of our prompting method, we create a dataset of 100 binary-choice questions, derived from an extensive schematic dataset on artifact parts and material composition. These questions ask which of two artifacts is less likely to share materials with another artifact. Such questions probe the LM's knowledge of shared materials in the part structure of different artifacts. We test our method on our dataset and three published commonsense reasoning datasets. The average accuracy of our method is consistently higher than that of all the other tested methods across all the tested datasets.

The game show Lucky 13 differs from other television game shows in that contestants are required to place a bet on their own knowledge of trivia by selecting a range that contains the number of questions that they answered correctly. We present a model for this game show using binomial random variables and generate tables outlining the optimal range the player should select based on maximization of two different utility functions. After analyzing the decisions made by some actual contestants on this show, we present a numerical simulation for how many questions an average player is expected to answer correctly based on question categories observed for two sample contestants.

We publicly release a new large-scale dataset, called SearchQA, for machine comprehension, or question-answering. Unlike recently released datasets, such as DeepMind CNN/DailyMail and SQuAD, the proposed SearchQA was constructed to reflect a full pipeline of general question-answering. That is, we start not from an existing article and generate a question-answer pair, but start from an existing question-answer pair, crawled from J! Archive, and augment it with text snippets retrieved by Google. Following this approach, we built SearchQA, which consists of more than 140k question-answer pairs with each pair having 49.6 snippets on average. Each question-answer-context tuple of the SearchQA comes with additional meta-data such as the snippet's URL, which we believe will be valuable resources for future research. We conduct human evaluation as well as test two baseline methods, one simple word selection and the other deep learning based, on the SearchQA. We show that there is a meaningful gap between the human and machine performances. This suggests that the proposed dataset could well serve as a benchmark for question-answering.

We propose a framework for robust evaluation of reasoning capabilities of language models, using functional variants of benchmarks. Models that solve a reasoning test should exhibit no difference in performance over the static version of a problem compared to a snapshot of the functional variant. We have rewritten the relevant fragment of the MATH benchmark into its functional variant MATH(), with functionalization of other benchmarks to follow. When evaluating current state-of-the-art models over snapshots of MATH(), we find a reasoning gap -- the percentage difference between the static and functional accuracies. We find reasoning gaps from 58.35% to 80.31% among the state-of-the-art closed and open weights models that perform well on static benchmarks, with the caveat that the gaps are likely to be smaller with more sophisticated prompting strategies. Here we show that models which anecdotally have good reasoning performance over real-world tasks, have quantifiable lower gaps, motivating the open problem of building "gap 0" models. Code for evaluation and new evaluation datasets, three MATH() snapshots, are publicly available at https://github.com/consequentai/fneval/.

Large Language Models (LLMs) have demonstrated remarkable capabilities across various applications, fundamentally reshaping the landscape of natural language processing (NLP) research. However, recent evaluation frameworks often rely on the output probabilities of LLMs for predictions, primarily due to computational constraints, diverging from real-world LLM usage scenarios. While widely employed, the efficacy of these probability-based evaluation strategies remains an open research question. This study aims to scrutinize the validity of such probability-based evaluation methods within the context of using LLMs for Multiple Choice Questions (MCQs), highlighting their inherent limitations. Our empirical investigation reveals that the prevalent probability-based evaluation method inadequately aligns with generation-based prediction. Furthermore, current evaluation frameworks typically assess LLMs through predictive tasks based on output probabilities rather than directly generating responses, owing to computational limitations. We illustrate that these probability-based approaches do not effectively correspond with generative predictions. The outcomes of our study can enhance the understanding of LLM evaluation methodologies and provide insights for future research in this domain.

Humans have a powerful and mysterious capacity to reason. By working through a series of purely mental steps, we can make inferences we would not be capable of making directly -- despite the fact that we get no additional data from the world. Similarly, when large language models generate a series of intermediate steps (a chain of thought) before answering a question, they often produce better answers than they otherwise would. We investigate why and how chain-of-thought reasoning is useful in language models, testing the hypothesis that reasoning is effective when training data consists of local clusters of variables that influence each other strongly. These training conditions enable the chaining of accurate local inferences in order to estimate relationships between variables that were not seen together in training. We prove that there will exist a "reasoning gap", where reasoning through intermediate variables improves inference, for the simple case of an autoregressive density estimator trained on local samples from a chain-structured probabilistic model. We then test our hypothesis empirically in more complex models, training an autoregressive language model on samples from Bayes nets but only including a subset of variables in each sample. We test language models' ability to match conditional probabilities with and without intermediate reasoning steps, finding that intermediate steps are only helpful when the training data is locally structured with respect to dependencies between variables and that the combination of locally-structured observations and reasoning is much more data-efficient than training on all variables. Our results illustrate how the effectiveness of reasoning step by step is rooted in the local statistical structure of the training data.

Sensitivity to false assumptions (or false premises) in information-seeking questions is critical for robust question-answering (QA) systems. Recent work has shown that false assumptions in naturally occurring questions pose challenges to current models, with low performance on both generative QA and simple detection tasks (Kim et al. 2023). However, the focus of existing work on naturally occurring questions leads to a gap in the analysis of model behavior on the long tail of the distribution of possible questions. To this end, we introduce Syn-(QA)^2, a set of two synthetically generated QA datasets: one generated using perturbed relations from Wikidata, and the other by perturbing HotpotQA (Yang et al. 2018). Our findings from evaluating a range of large language models are threefold: (1) false assumptions in QA are challenging, echoing the findings of prior work, (2) the binary detection task is challenging even compared to the difficulty of generative QA itself, possibly due to the linguistic structure of the problem, and (3) the detection task is more challenging with long-tail questions compared to naturally occurring questions, highlighting the utility of our synthetic datasets and generation method.

We conduct a feasibility study into the applicability of answer-agnostic question generation models to textbook passages. We show that a significant portion of errors in such systems arise from asking irrelevant or uninterpretable questions and that such errors can be ameliorated by providing summarized input. We find that giving these models human-written summaries instead of the original text results in a significant increase in acceptability of generated questions (33% rightarrow 83%) as determined by expert annotators. We also find that, in the absence of human-written summaries, automatic summarization can serve as a good middle ground.

We present a simple method that achieves unexpectedly superior performance for Complex Reasoning involved Visual Question Answering. Our solution collects statistical features from high-frequency words of all the questions asked about an image and use them as accurate knowledge for answering further questions of the same image. We are fully aware that this setting is not ubiquitously applicable, and in a more common setting one should assume the questions are asked separately and they cannot be gathered to obtain a knowledge base. Nonetheless, we use this method as an evidence to demonstrate our observation that the bottleneck effect is more severe on the feature extraction part than it is on the knowledge reasoning part. We show significant gaps when using the same reasoning model with 1) ground-truth features; 2) statistical features; 3) detected features from completely learned detectors, and analyze what these gaps mean to researches on visual reasoning topics. Our model with the statistical features achieves the 2nd place in the GQA Challenge 2019.

Current benchmarks for AI clinician systems, often based on multiple-choice exams or manual rubrics, fail to capture the depth, robustness, and safety required for real-world clinical practice. To address this, we introduce the GAPS framework, a multidimensional paradigm for evaluating Grounding (cognitive depth), Adequacy (answer completeness), Perturbation (robustness), and Safety. Critically, we developed a fully automated, guideline-anchored pipeline to construct a GAPS-aligned benchmark end-to-end, overcoming the scalability and subjectivity limitations of prior work. Our pipeline assembles an evidence neighborhood, creates dual graph and tree representations, and automatically generates questions across G-levels. Rubrics are synthesized by a DeepResearch agent that mimics GRADE-consistent, PICO-driven evidence review in a ReAct loop. Scoring is performed by an ensemble of large language model (LLM) judges. Validation confirmed our automated questions are high-quality and align with clinician judgment. Evaluating state-of-the-art models on the benchmark revealed key failure modes: performance degrades sharply with increased reasoning depth (G-axis), models struggle with answer completeness (A-axis), and they are highly vulnerable to adversarial perturbations (P-axis) as well as certain safety issues (S-axis). This automated, clinically-grounded approach provides a reproducible and scalable method for rigorously evaluating AI clinician systems and guiding their development toward safer, more reliable clinical practice.

Large language models can arrive at the same answer through reasoning paths that are unstable, contradictory, or difficult to rank consistently -- a failure mode especially prevalent in multi-step deductive reasoning. Existing methods assess reliability primarily through output dispersion -- measuring how much sampled answers differ -- but this discards a complementary signal: whether the model can consistently rank competing reasoning candidates. We propose structural uncertainty, a consistency-aware framework derived from the stability of self-preference-induced rankings over sampled reasoning solutions. Given a query, we generate multiple candidate solutions and ask the model to judge pairwise preferences among its own outputs. We aggregate self-preferences into ranking distributions via Bradley-Terry modeling with PageRank, and decompose the signal into two entropy-based components: across-trial ranking instability and within-trial candidate ambiguity. Across five LLMs and eight benchmarks, structural signals provide information complementary to answer dispersion: on logical and mathematical reasoning tasks, the combination improves identification of unreliable instances, while on factual retrieval the structural signal collapses toward uniformity, diagnosing a regime boundary where reasoning-level consistency evaluation is uninformative. The two components relate differently to accuracy: within-trial ambiguity correlates positively with correctness -- consistent with settings where multiple plausible solution paths remain competitive -- while across-trial instability correlates negatively, signaling unreliable reasoning. Structural uncertainty is best understood not as a universal confidence estimator, but as a regime-sensitive evaluator of logical reasoning consistency.

Multi-hop question answering (MHQA) requires a model to retrieve and integrate information from multiple passages to answer a complex question. Recent systems leverage the power of large language models and integrate evidence retrieval with reasoning prompts (e.g., chain-of-thought reasoning) for the MHQA task. However, the complexities in the question types (bridge v.s. comparison questions) and the reasoning types (sequential v.s. parallel reasonings) require more novel and fine-grained prompting methods to enhance the performance of MHQA under the zero-shot setting. In this paper, we propose STOC-TOT, a stochastic tree-of-thought reasoning prompting method with constrained decoding for MHQA and conduct a detailed comparison with other reasoning prompts on different question types and reasoning types. Specifically, we construct a tree-like reasoning structure by prompting the model to break down the original question into smaller sub-questions to form different reasoning paths. In addition, we prompt the model to provide a probability estimation for each reasoning path at each reasoning step. At answer time, we conduct constrained decoding on the model to generate more grounded answers and reduce hallucination. Experiments comparing STOC-TOT with two MHQA datasets and five large language models showed that our framework outperforms other reasoning prompts by a significant margin.

This paper presents a case study on the design, administration, post-processing, and evaluation of surveys on large language models (LLMs). It comprises two components: (1) A statistical method for eliciting beliefs encoded in LLMs. We introduce statistical measures and evaluation metrics that quantify the probability of an LLM "making a choice", the associated uncertainty, and the consistency of that choice. (2) We apply this method to study what moral beliefs are encoded in different LLMs, especially in ambiguous cases where the right choice is not obvious. We design a large-scale survey comprising 680 high-ambiguity moral scenarios (e.g., "Should I tell a white lie?") and 687 low-ambiguity moral scenarios (e.g., "Should I stop for a pedestrian on the road?"). Each scenario includes a description, two possible actions, and auxiliary labels indicating violated rules (e.g., "do not kill"). We administer the survey to 28 open- and closed-source LLMs. We find that (a) in unambiguous scenarios, most models "choose" actions that align with commonsense. In ambiguous cases, most models express uncertainty. (b) Some models are uncertain about choosing the commonsense action because their responses are sensitive to the question-wording. (c) Some models reflect clear preferences in ambiguous scenarios. Specifically, closed-source models tend to agree with each other.

Multiple choice questions (MCQs) are commonly used to evaluate the capabilities of large language models (LLMs). One common way to evaluate the model response is to rank the candidate answers based on the log probability of the first token prediction. An alternative way is to examine the text output. Prior work has shown that first token probabilities lack robustness to changes in MCQ phrasing, and that first token probabilities do not match text answers for instruction-tuned models. Therefore, in this paper, we investigate the robustness of text answers. We show that the text answers are more robust to question perturbations than the first token probabilities, when the first token answers mismatch the text answers. The difference in robustness increases as the mismatch rate becomes greater. As the mismatch reaches over 50\%, the text answer is more robust to option order changes than the debiased first token probabilities using state-of-the-art debiasing methods such as PriDe. Our findings provide further evidence for the benefits of text answer evaluation over first token probability evaluation.

One of the most widely used methods to evaluate LLMs are Multiple Choice Question (MCQ) tests. MCQ benchmarks enable the testing of LLM knowledge on almost any topic at scale as the results can be processed automatically. To help the LLM answer, a few examples called few shots can be included in the prompt. Moreover, the LLM can be asked to answer the question directly with the selected option or to first provide the reasoning and then the selected answer, which is known as chain of thought. In addition to checking whether the selected answer is correct, the evaluation can look at the LLM-estimated probability of its response as an indication of the confidence of the LLM in the response. In this paper, we study how the LLM confidence in its answer depends on whether the model has been asked to answer directly or to provide the reasoning before answering. The results of the evaluation of questions on a wide range of topics in seven different models show that LLMs are more confident in their answers when they provide reasoning before the answer. This occurs regardless of whether the selected answer is correct. Our hypothesis is that this behavior is due to the reasoning that modifies the probability of the selected answer, as the LLM predicts the answer based on the input question and the reasoning that supports the selection made. Therefore, LLM estimated probabilities seem to have intrinsic limitations that should be understood in order to use them in evaluation procedures. Interestingly, the same behavior has been observed in humans, for whom explaining an answer increases confidence in its correctness.

A benchmark score means nothing without knowing what a trivial method achieves and what the best possible method could achieve. We construct both bounds for a task with a rare kind of ground truth: predicting which sentences a crowd of readers -- highlighting for their own purposes, unpaid, uninstructed, and blind to each other -- marked in 120 web documents. The floor is naive truncation (lead); the ceiling is a split-half oracle: half the crowd predicting the other half. The gap between them is +0.2028 AP [+0.1698, +0.2342, domain-clustered], and three findings structure it. First, the gap is semantic: position and length features recover 5% of it. Second, frontier language models reach 35-53% of it zero-shot -- far above classical baselines, far below the crowd; a state-of-the-art prompt compressor (LLMLingua-2) lands below the floor, indistinguishable from random selection. Third, an unweighted cross-vendor fusion of five frontier rankings plus a position prior reaches 60%, beating the best single model by +0.0159 [+0.0044, +0.0269; Holm p=0.019] -- a gain that survives ablation of its best member, split-half arm selection, prompt paraphrase, and label, gate, and seed perturbations, and was CONFIRMED by a pre-registered replication on 217 independent documents (+0.0179, Holm p=0.042). Finally, the bracket compresses: distilling the fusion into one open-weight 8B student that reads the whole document retains 90% of the fusion's edge and reaches statistical parity with the strongest single frontier model (+0.0070 [-0.0068, +0.0200]), where a local-context student retains only 63% -- the crowd's signal lives in document-level structure, and the cheapest known improvement is to ask several different models and average.

Large language models (LLMs) exhibit positional bias in how they use context, which especially complicates listwise ranking. To address this, we propose permutation self-consistency, a form of self-consistency over ranking list outputs of black-box LLMs. Our key idea is to marginalize out different list orders in the prompt to produce an order-independent ranking with less positional bias. First, given some input prompt, we repeatedly shuffle the list in the prompt and pass it through the LLM while holding the instructions the same. Next, we aggregate the resulting sample of rankings by computing the central ranking closest in distance to all of them, marginalizing out prompt order biases in the process. Theoretically, we prove the robustness of our method, showing convergence to the true ranking in the presence of random perturbations. Empirically, on five list-ranking datasets in sorting and passage reranking, our approach improves scores from conventional inference by up to 7-18% for GPT-3.5 and 8-16% for LLaMA v2 (70B), surpassing the previous state of the art in passage reranking. Our code is at https://github.com/castorini/perm-sc.

Words of estimative probability (WEP) are expressions of a statement's plausibility (probably, maybe, likely, doubt, likely, unlikely, impossible...). Multiple surveys demonstrate the agreement of human evaluators when assigning numerical probability levels to WEP. For example, highly likely corresponds to a median chance of 0.90+-0.08 in Fagen-Ulmschneider (2015)'s survey. In this work, we measure the ability of neural language processing models to capture the consensual probability level associated to each WEP. Firstly, we use the UNLI dataset (Chen et al., 2020) which associates premises and hypotheses with their perceived joint probability p, to construct prompts, e.g. "[PREMISE]. [WEP], [HYPOTHESIS]." and assess whether language models can predict whether the WEP consensual probability level is close to p. Secondly, we construct a dataset of WEP-based probabilistic reasoning, to test whether language models can reason with WEP compositions. When prompted "[EVENTA] is likely. [EVENTB] is impossible.", a causal language model should not express that [EVENTA&B] is likely. We show that both tasks are unsolved by off-the-shelf English language models, but that fine-tuning leads to transferable improvement.

Multiple-choice benchmarks are widely used to evaluate large language models, but MCQ scores conflate knowledge with sensitivity to option order, which makes them unreliable measures of model knowledge. In this paper, we test whether preventing a model from seeing option labels while committing to an answer removes positional influence and, in turn, improves performance. We evaluate two different strategies for mitigating bias. The first uses a generation-then-matching approach, and the second scores options in isolation, which is positionally unbiased by construction. Neither reliably improves accuracy. A complete decomposition shows that the bottleneck is withholding options, not the matching step. The only configuration that consistently matches the baseline is the one that shows the model all options paired with an LLM matcher. However, eliminating positional influence entirely still does not reliably yield accuracy gains, while cyclic permutation often improves them. For two-stage prompting, an aggregate measure of recall imbalance and a direct per-question measure of order sensitivity both fail to show reliable debiasing.

As we embark on a new era of LLMs, it becomes increasingly crucial to understand their capabilities, limitations, and differences. Toward making further progress in this direction, we strive to build a deeper understanding of the gaps between massive LLMs (e.g., ChatGPT) and smaller yet effective open-source LLMs and their distilled counterparts. To this end, we specifically focus on long-form question answering (LFQA) because it has several practical and impactful applications (e.g., troubleshooting, customer service, etc.) yet is still understudied and challenging for LLMs. We propose a question-generation method from abstractive summaries and show that generating follow-up questions from summaries of long documents can create a challenging setting for LLMs to reason and infer from long contexts. Our experimental results confirm that: (1) our proposed method of generating questions from abstractive summaries pose a challenging setup for LLMs and shows performance gaps between LLMs like ChatGPT and open-source LLMs (Alpaca, Llama) (2) open-source LLMs exhibit decreased reliance on context for generated questions from the original document, but their generation capabilities drop significantly on generated questions from summaries -- especially for longer contexts (>1024 tokens)

The rapid development of LLMs has sparked extensive research into their factual knowledge. Current works find that LLMs fall short on questions around low-frequency entities. However, such proofs are unreliable since the questions can differ not only in entity frequency but also in difficulty themselves. So we introduce ComparisonQA benchmark, containing 283K abstract questions, each instantiated by a pair of high-frequency and low-frequency entities. It ensures a controllable comparison to study the role of knowledge frequency in the performance of LLMs. Because the difference between such a pair is only the entity with different frequencies. In addition, we use both correctness and uncertainty to develop a two-round method to evaluate LLMs' knowledge robustness. It aims to avoid possible semantic shortcuts which is a serious problem of current QA study. Experiments reveal that LLMs, including GPT-4o, exhibit particularly low robustness regarding low-frequency knowledge. Besides, we find that uncertainty can be used to effectively identify high-quality and shortcut-free questions while maintaining the data size. Based on this, we propose an automatic method to select such questions to form a subset called ComparisonQA-Hard, containing only hard low-frequency questions.

Many modern AI question-answering systems convert text into vectors and retrieve the closest matches to a user question. While effective for topical similarity, similarity scores alone do not explain why some retrieved text can serve as evidence while other equally similar text cannot. When many candidates receive similar scores, systems may select sentences that are redundant, incomplete, or address different conditions than the question requires. This paper presents a deterministic evidence selection framework for retrieval-augmented question answering. The approach introduces Meaning-Utility Estimation (MUE) and Diversity-Utility Estimation (DUE), fixed scoring and redundancy-control procedures that determine evidence admissibility prior to answer generation. Each sentence or record is evaluated independently using explicit signals for semantic relatedness, term coverage, conceptual distinctiveness, and redundancy. No training or fine-tuning is required. In the prototype, a unit is accepted only if it explicitly states the fact, rule, or condition required by the task. Units are not merged or expanded. If no unit independently satisfies the requirement, the system returns no answer. This deterministic gating produces compact, auditable evidence sets and establishes a clear boundary between relevant text and usable evidence.

Survey paper plays a crucial role in scientific research, especially given the rapid growth of research publications. Recently, researchers have begun using LLMs to automate survey generation for better efficiency. However, the quality gap between LLM-generated surveys and those written by human remains significant, particularly in terms of outline quality and citation accuracy. To close these gaps, we introduce SurveyForge, which first generates the outline by analyzing the logical structure of human-written outlines and referring to the retrieved domain-related articles. Subsequently, leveraging high-quality papers retrieved from memory by our scholar navigation agent, SurveyForge can automatically generate and refine the content of the generated article. Moreover, to achieve a comprehensive evaluation, we construct SurveyBench, which includes 100 human-written survey papers for win-rate comparison and assesses AI-generated survey papers across three dimensions: reference, outline, and content quality. Experiments demonstrate that SurveyForge can outperform previous works such as AutoSurvey.

We present a new probing dataset named PROST: Physical Reasoning about Objects Through Space and Time. This dataset contains 18,736 multiple-choice questions made from 14 manually curated templates, covering 10 physical reasoning concepts. All questions are designed to probe both causal and masked language models in a zero-shot setting. We conduct an extensive analysis which demonstrates that state-of-the-art pretrained models are inadequate at physical reasoning: they are influenced by the order in which answer options are presented to them, they struggle when the superlative in a question is inverted (e.g., most <-> least), and increasing the amount of pretraining data and parameters only yields minimal improvements. These results provide support for the hypothesis that current pretrained models' ability to reason about physical interactions is inherently limited by a lack of real world experience. By highlighting these limitations, we hope to motivate the development of models with a human-like understanding of the physical world.

Estimating the difficulty of multiple-choice questions would be great help for educators who must spend substantial time creating and piloting stimuli for their tests, and for learners who want to practice. Supervised approaches to difficulty estimation have yielded to date mixed results. In this contribution we leverage an aspect of generative large models which might be seen as a weakness when answering questions, namely their uncertainty, and exploit it towards exploring correlations between two different metrics of uncertainty, and the actual student response distribution. While we observe some present but weak correlations, we also discover that the models' behaviour is different in the case of correct vs wrong answers, and that correlations differ substantially according to the different question types which are included in our fine-grained, previously unused dataset of 451 questions from a Biopsychology course. In discussing our findings, we also suggest potential avenues to further leverage model uncertainty as an additional proxy for item difficulty.

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.

Despite the strong performance of large language models (LLMs) in tasks like mathematical reasoning, their practical use is limited by high computational demands and proprietary restrictions. Chain-of-thought (CoT) and program-of-thought (PoT) fine-tuning are common methods to transfer LLM knowledge to small language models (SLMs). However, CoT often leads to calculation errors in SLMs, while PoT has shown more promise. While most PoT-based approaches focus on direct problem-to-code conversion or extracting only the key information from questions and then providing code solution for it, this work emphasizes filling the gaps in the question to clearly illustrate the solution path, which can be challenging for an SLM to understand when such information is not explicitly provided. Therefore, this paper introduces Gap-Filling Prompting (GFP), a novel two-step prompting strategy designed to enhance the problem-solving process for SLMs. The first step identifies these gaps and provides hints for filling them, while the second step adds the hints to the question to generate a final code solution. Experimental results on two benchmark datasets demonstrate that GFP significantly improves the mathematical reasoning abilities of SLMs.

This study investigates whether repeating questions within prompts influences the performance of large language models (LLMs). We hypothesize that reiterating a question within a single prompt might enhance the model's focus on key elements of the query. We evaluate five recent LLMs -- including GPT-4o-mini, DeepSeek-V3, and smaller open-source models -- on three reading comprehension datasets under different prompt settings, varying question repetition levels (1, 3, or 5 times per prompt). Our results demonstrate that question repetition can increase models' accuracy by up to 6%. However, across all models, settings, and datasets, we do not find the result statistically significant. These findings provide insights into prompt design and LLM behavior, suggesting that repetition alone does not significantly impact output quality.

Demographic synthetic survey panels are often validated by matching aggregate answers to published surveys. We test what that certificate establishes across six multiselect batteries from four survey organisations in three countries. The headline analysis is restricted to three instruments whose synthetic cohort and human target share the stated population frame; three other batteries remain sensitivity analyses. The response contract dominates measured fidelity. In the aligned instruments, committed sets leave 66 of 128 model-battery option slots empty in panels of up to 500 respondents, versus 0 of 128 under per-option probability elicitation. Across eight uncapped model-instrument comparisons, probabilities reduce option-marginal MAE by 4.53 to 7.30 points. The capped instrument reverses on two models until the vectors are projected onto its stated maximum. These are measurement effects: human targets are realised check-all responses, whereas the vectors are latent inclusion propensities. Published marginal agreement also fails to discriminate respondent simulation from direct population estimation. On nine aligned model-battery pairs, a no-persona population-prevalence query averages 6.27 MAE versus 12.39 for committed panels and wins all nine comparisons. Constraint-aware probability vectors average 5.34 and beat the query on four of nine, so the baseline challenges the validation criterion rather than proving direct estimation uniformly best. On three unpublished demographic cells, neither approach beats reciting the national distribution. Population-marginal agreement is therefore evidence about an elicitation contract and an estimand obtainable without simulated respondents, not evidence of individual simulation.

We curate a comprehensive dataset of 4,550 questions and solutions from problem sets, midterm exams, and final exams across all MIT Mathematics and Electrical Engineering and Computer Science (EECS) courses required for obtaining a degree. We evaluate the ability of large language models to fulfill the graduation requirements for any MIT major in Mathematics and EECS. Our results demonstrate that GPT-3.5 successfully solves a third of the entire MIT curriculum, while GPT-4, with prompt engineering, achieves a perfect solve rate on a test set excluding questions based on images. We fine-tune an open-source large language model on this dataset. We employ GPT-4 to automatically grade model responses, providing a detailed performance breakdown by course, question, and answer type. By embedding questions in a low-dimensional space, we explore the relationships between questions, topics, and classes and discover which questions and classes are required for solving other questions and classes through few-shot learning. Our analysis offers valuable insights into course prerequisites and curriculum design, highlighting language models' potential for learning and improving Mathematics and EECS education.

In recent years, corpus-driven AI methods, such as Large Language Models (LLMs), have seen widespread use in education. While on the surface their abilities look promising for tasks ranging from generating assessment materials to simulating student performance, we should be aware of the subtle nuances of their frequentist nature that might be affecting their behaviour. In this work, we focus on the aspect of corpus frequency in the context of creating high-quality Multiple Choice Questions (MCQs), specifically asking: What if corpus prevalence were enough to identify the correct answer to an MCQ? We propose a computational method of assessing corpus prevalence of MCQ options in large text corpora leveraging textual embeddings using both expert- and machine-generated MCQ sets. The key finding, across three large question sets, is that correct answers, independently of the question stem, are significantly more available than incorrect options. Specifically, using Wikipedia as the retrieval corpus, we find that always selecting the most prevalent option leads to scores up to 9.0% above the random-guess baseline. We also find that MCQ distractors generated by LLMs often show similar patterns of prevalence compared to expert-created options, despite the LLMs' frequentist nature and their training on large collections of textual data. Moreover, we find that corpus prevalence does not necessarily correlate with how recognisable terms are to humans. This highlights the need to better understand how corpora are used in AI-driven methods for education, whether applied directly or indirectly via LLMs.

Readers of academic research papers often read with the goal of answering specific questions. Question Answering systems that can answer those questions can make consumption of the content much more efficient. However, building such tools requires data that reflect the difficulty of the task arising from complex reasoning about claims made in multiple parts of a paper. In contrast, existing information-seeking question answering datasets usually contain questions about generic factoid-type information. We therefore present QASPER, a dataset of 5,049 questions over 1,585 Natural Language Processing papers. Each question is written by an NLP practitioner who read only the title and abstract of the corresponding paper, and the question seeks information present in the full text. The questions are then answered by a separate set of NLP practitioners who also provide supporting evidence to answers. We find that existing models that do well on other QA tasks do not perform well on answering these questions, underperforming humans by at least 27 F1 points when answering them from entire papers, motivating further research in document-grounded, information-seeking QA, which our dataset is designed to facilitate.

This paper examines the zero-shot ability of Large Language Models (LLMs) to detect multiple-choice questions with no correct answer, a crucial aspect of educational assessment quality. We explore this ability not only as a measure of subject matter knowledge but also as an indicator of critical thinking within LLMs. Our experiments, utilizing a range of LLMs on diverse questions, highlight the significant performance gap between questions with a single correct answer and those without. Llama-3.1-405B stands out by successfully identifying the lack of a valid answer in many instances. These findings suggest that LLMs should prioritize critical thinking over blind instruction following and caution against their use in educational settings where questions with incorrect answers might lead to inaccurate evaluations. This research sets a benchmark for assessing critical thinking in LLMs and emphasizes the need for ongoing model alignment to ensure genuine user comprehension and assistance.

Accurately quantifying a large language model's (LLM) predictive uncertainty is crucial for judging the reliability of its answers. While most existing research focuses on short, directly answerable questions with closed-form outputs (e.g., multiple-choice), involving intermediate reasoning steps in LLM responses is increasingly important. This added complexity complicates uncertainty quantification (UQ) because the probabilities assigned to answer tokens are conditioned on a vast space of preceding reasoning tokens. Direct marginalization is infeasible, and the dependency inflates probability estimates, causing overconfidence in UQ. To address this, we propose UQAC, an efficient method that narrows the reasoning space to a tractable size for marginalization. UQAC iteratively constructs an "attention chain" of tokens deemed "semantically crucial" to the final answer via a backtracking procedure. Starting from the answer tokens, it uses attention weights to identify the most influential predecessors, then iterates this process until reaching the input tokens. Similarity filtering and probability thresholding further refine the resulting chain, allowing us to approximate the marginal probabilities of the answer tokens, which serve as the LLM's confidence. We validate UQAC on multiple reasoning benchmarks with advanced open-source LLMs, demonstrating that it consistently delivers reliable UQ estimates with high computational efficiency.

Large Language Models (LLMs) have demonstrated remarkable capabilities in various educational tasks, yet their alignment with human learning patterns, particularly in predicting which incorrect options students are most likely to select in multiple-choice questions (MCQs), remains underexplored. Our work investigates the relationship between LLM generation likelihood and student response distributions in MCQs with a specific focus on distractor selections. We collect a comprehensive dataset of MCQs with real-world student response distributions to explore two fundamental research questions: (1). RQ1 - Do the distractors that students more frequently select correspond to those that LLMs assign higher generation likelihood to? (2). RQ2 - When an LLM selects a incorrect choice, does it choose the same distractor that most students pick? Our experiments reveals moderate correlations between LLM-assigned probabilities and student selection patterns for distractors in MCQs. Additionally, when LLMs make mistakes, they are more likley to select the same incorrect answers that commonly mislead students, which is a pattern consistent across both small and large language models. Our work provides empirical evidence that despite LLMs' strong performance on generating educational content, there remains a gap between LLM's underlying reasoning process and human cognitive processes in identifying confusing distractors. Our findings also have significant implications for educational assessment development. The smaller language models could be efficiently utilized for automated distractor generation as they demonstrate similar patterns in identifying confusing answer choices as larger language models. This observed alignment between LLMs and student misconception patterns opens new opportunities for generating high-quality distractors that complement traditional human-designed distractors.

A key limitation in current datasets for multi-hop reasoning is that the required steps for answering the question are mentioned in it explicitly. In this work, we introduce StrategyQA, a question answering (QA) benchmark where the required reasoning steps are implicit in the question, and should be inferred using a strategy. A fundamental challenge in this setup is how to elicit such creative questions from crowdsourcing workers, while covering a broad range of potential strategies. We propose a data collection procedure that combines term-based priming to inspire annotators, careful control over the annotator population, and adversarial filtering for eliminating reasoning shortcuts. Moreover, we annotate each question with (1) a decomposition into reasoning steps for answering it, and (2) Wikipedia paragraphs that contain the answers to each step. Overall, StrategyQA includes 2,780 examples, each consisting of a strategy question, its decomposition, and evidence paragraphs. Analysis shows that questions in StrategyQA are short, topic-diverse, and cover a wide range of strategies. Empirically, we show that humans perform well (87%) on this task, while our best baseline reaches an accuracy of sim66%.

Reasoning quality in large language models depends not only on producing correct answers but also on generating valid intermediate steps. We study this through multiple-choice question answering (MCQA), which provides a controlled setting with fixed answer options. Our analysis shows that when questions are effectively unsolvable for a model, spurious chains of thought (CoTs) are more likely to appear, leading to false positives. By estimating the solvability of each question, we uncover an intermediate regime where learning is most effective. Building on this insight, we adapt outcome-supervised reward models and reinforcement learning with group-relative advantage to incorporate solvability into their objectives. Across experiments on math and multimodal datasets, these modifications consistently yield higher rates of process-correct reasoning and, in reinforcement learning, improved answer accuracy as well. Our results highlight solvability as a key factor for reducing hallucinations and increasing reliability in CoT reasoning.

Recent studies show that large language models (LLMs) can be instructed to effectively perform zero-shot passage re-ranking, in which the results of a first stage retrieval method, such as BM25, are rated and reordered to improve relevance. In this work, we improve LLM-based re-ranking by algorithmically selecting few-shot demonstrations to include in the prompt. Our analysis investigates the conditions where demonstrations are most helpful, and shows that adding even one demonstration is significantly beneficial. We propose a novel demonstration selection strategy based on difficulty rather than the commonly used semantic similarity. Furthermore, we find that demonstrations helpful for ranking are also effective at question generation. We hope our work will spur more principled research into question generation and passage ranking.

Large language models (LLMs) have shown impressive performance in various reasoning benchmarks with the emergence of Chain-of-Thought (CoT) and its derivative methods, particularly in tasks involving multi-choice questions (MCQs). However, current works all process data uniformly without considering the problem-solving difficulty, which means an excessive focus on simple questions while insufficient to intricate ones. To address this challenge, we inspired by humans using heuristic strategies to categorize tasks and handle them individually, propose to apply the Divide and Conquer to LLMs reasoning. First, we divide questions into different subsets based on the statistical confidence score (CS), then fix nearly resolved sets and conquer demanding nuanced process ones with elaborately designed methods, including Prior Knowledge based Reasoning (PKR) and Filter Choices based Reasoning (FCR), as well as their integration variants. Our experiments demonstrate that this proposed strategy significantly boosts the models' reasoning abilities across nine datasets involving arithmetic, commonsense, and logic tasks. For instance, compared to baseline, we make a striking improvement on low confidence subsets of 8.72\% for AQuA, 15.07\% for ARC Challenge and 7.71\% for RiddleSense. In addition, through extensive analysis on length of rationale and number of options, we verify that longer reasoning paths in PKR could prevent models from referring infer-harmful shortcuts, and also find that removing irrelevant choices in FCR would substantially avoid models' confusion. The code is at https://github.com/AiMijie/Divide-and-Conquer

Between April 1 and May 15, 2026, a group of 49 mathematicians compiled a dataset of research-level mathematics questions with known answers. Most of the work was done during the 3-day workshop *Benchmarks in Leipzig* with 35 participants at the Max Planck Institute for Mathematics in the Sciences in Leipzig, Germany. We present the resulting collection of 100 questions. We evaluated these questions in three stages: a single attempt by five state-of-the-art LLMs, followed by a 20-runs-per-model evaluation with three of these models, and finally a 3-run attempt with two heavy-thinking models. After Stage 1, 41 questions remained completely unsolved; after Stage 2, this count dropped to 16; and we concluded Stage 3 with only 2 unsolved questions. This demonstrates that the mathematical reasoning capabilities of LLMs are becoming impressive.

We introduce Browsing Lost Unformed Recollections, a tip-of-the-tongue known-item search and reasoning benchmark for general AI assistants. BLUR introduces a set of 573 real-world validated questions that demand searching and reasoning across multi-modal and multilingual inputs, as well as proficient tool use, in order to excel on. Humans easily ace these questions (scoring on average 98%), while the best-performing system scores around 56%. To facilitate progress toward addressing this challenging and aspirational use case for general AI assistants, we release 350 questions through a public leaderboard, retain the answers to 250 of them, and have the rest as a private test set.

Language models (LM) are capable of remarkably complex linguistic tasks; however, numerical reasoning is an area in which they frequently struggle. An important but rarely evaluated form of reasoning is understanding probability distributions. In this paper, we focus on evaluating the probabilistic reasoning capabilities of LMs using idealized and real-world statistical distributions. We perform a systematic evaluation of state-of-the-art LMs on three tasks: estimating percentiles, drawing samples, and calculating probabilities. We evaluate three ways to provide context to LMs 1) anchoring examples from within a distribution or family of distributions, 2) real-world context, 3) summary statistics on which to base a Normal approximation. Models can make inferences about distributions, and can be further aided by the incorporation of real-world context, example shots and simplified assumptions, even if these assumptions are incorrect or misspecified. To conduct this work, we developed a comprehensive benchmark distribution dataset with associated question-answer pairs that we will release publicly.

Large Language Models (LLMs) are revolutionizing information retrieval, with chatbots becoming an important source for answering user queries. As by their design, LLMs prioritize generating correct answers, the value of highly plausible yet incorrect answers (candidate answers) tends to be overlooked. However, such answers can still prove useful, for example, they can play a crucial role in tasks like Multiple-Choice Question Answering (MCQA) and QA Robustness Assessment (QARA). Existing QA datasets primarily focus on correct answers without explicit consideration of the plausibility of other candidate answers, limiting opportunity for more nuanced evaluations of models. To address this gap, we introduce PlausibleQA, a large-scale dataset comprising 10,000 questions and 100,000 candidate answers, each annotated with plausibility scores and justifications for their selection. Additionally, the dataset includes 900,000 justifications for pairwise comparisons between candidate answers, further refining plausibility assessments. We evaluate PlausibleQA through human assessments and empirical experiments, demonstrating its utility in MCQA and QARA analysis. Our findings show that plausibility-aware approaches are effective for MCQA distractor generation and QARA. We release PlausibleQA as a resource for advancing QA research and enhancing LLM performance in distinguishing plausible distractors from correct answers.

Recent advancements in large language models (LLMs) often rely on generating intermediate reasoning steps to enhance accuracy. However, little work has examined how reasoning utility contributes to the final answer's correctness. Due to the stochastic nature of autoregressive generation, generating more context does not guarantee increased confidence in the answer. If we could predict, during generation, whether a reasoning step will be useful, we could stop early or prune ineffective steps, avoiding distractions in the final decision. We present an oracle study on MATH dataset, using Qwen2.5-32B and GPT-4o to generate reasoning chains, and then employing a separate model (Qwen3-8B) to quantify the utility of these chains for final accuracy. Specifically, we measure the model's uncertainty on the answer span Y at each reasoning step using conditional entropy (expected negative log-likelihood over the vocabulary) with context expanding step by step. Our results show a clear pattern: conditional entropy that decreases over steps is strongly associated with correct answers, whereas flat or increasing entropy often results in wrong answers. We also corroborate that incorrect reasoning paths tend to be longer than correct ones, suggesting that longer reasoning does not necessarily yield better outcomes. These findings serve as a foundation to inspire future work on designing efficient reasoning pipelines that detect and avoid unproductive reasoning early.

Information seeking users often pose questions with false presuppositions, especially when asking about unfamiliar topics. Most existing question answering (QA) datasets, in contrast, assume all questions have well defined answers. We introduce CREPE, a QA dataset containing a natural distribution of presupposition failures from online information-seeking forums. We find that 25% of questions contain false presuppositions, and provide annotations for these presuppositions and their corrections. Through extensive baseline experiments, we show that adaptations of existing open-domain QA models can find presuppositions moderately well, but struggle when predicting whether a presupposition is factually correct. This is in large part due to difficulty in retrieving relevant evidence passages from a large text corpus. CREPE provides a benchmark to study question answering in the wild, and our analyses provide avenues for future work in better modeling and further studying the task.

We propose a simple and effective re-ranking method for improving passage retrieval in open question answering. The re-ranker re-scores retrieved passages with a zero-shot question generation model, which uses a pre-trained language model to compute the probability of the input question conditioned on a retrieved passage. This approach can be applied on top of any retrieval method (e.g. neural or keyword-based), does not require any domain- or task-specific training (and therefore is expected to generalize better to data distribution shifts), and provides rich cross-attention between query and passage (i.e. it must explain every token in the question). When evaluated on a number of open-domain retrieval datasets, our re-ranker improves strong unsupervised retrieval models by 6%-18% absolute and strong supervised models by up to 12% in terms of top-20 passage retrieval accuracy. We also obtain new state-of-the-art results on full open-domain question answering by simply adding the new re-ranker to existing models with no further changes.

Evaluating long-form responses to research queries heavily relies on expert annotators, restricting attention to areas like AI where researchers can conveniently enlist colleagues. Yet, research expertise is widespread: survey articles synthesize knowledge distributed across the literature. We introduce ResearchQA, a resource for evaluating LLM systems by distilling survey articles from 75 research fields into 21K queries and 160K rubric items. Each rubric, derived jointly with queries from survey sections, lists query-specific answer evaluation criteria, i.e., citing papers, making explanations, and describing limitations. Assessments by 31 Ph.D. annotators in 8 fields indicate 96% of queries support Ph.D. information needs and 87% of rubric items should be addressed in system responses by a sentence or more. Using our rubrics, we are able to construct an automatic pairwise judge obtaining 74% agreement with expert judgments. We leverage ResearchQA to analyze competency gaps in 18 systems in over 7.6K pairwise evaluations. No parametric or retrieval-augmented system we evaluate exceeds 70% on covering rubric items, and the highest-ranking agentic system shows 75% coverage. Error analysis reveals that the highest-ranking system fully addresses less than 11% of citation rubric items, 48% of limitation items, and 49% of comparison items. We release our data to facilitate more comprehensive multi-field evaluations.

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%).

AI-search evaluation commonly treats a prompt as a stable query that can be counted, classified, and replayed in isolation. A conversation makes that unit of analysis questionable: each user turn can add a constraint, revise an assumption, request evidence, or refer to alternatives established earlier. We replace latent "intent" with an observable construct, conversation-conditioned request state, and measure how that state is distributed across user turns. The analysis reuses frozen rules and the governed cohort of a preceding conversation study: 670 English commercial multi-turn conversations in a discovery-replication design and 7,463 public PRISM conversations from 1,389 participants. In the commercial corpus, the final prompt contains a median 35.6% of the session's unique user-side content vocabulary; in PRISM, the median is 36.4%. The final prompt contains at most half of that vocabulary in 68.4% and 74.3% of conversations, respectively. More importantly, transparent rules detect at least one request-state dimension in history but not in the final prompt in 50.3% of commercial conversations and 44.8% of PRISM conversations. Among dimension-bearing conversations, the final prompt reproduces the full observed dimension set in only 26.1% and 26.2%. At the same time, the final prompt adds a previously unseen dimension in 17.9% and 19.3%, showing that the endpoint is neither a summary nor merely a reference: it is often another state update. Length-matched nulls show that low lexical coverage is largely a consequence of turn length, so vocabulary results are interpreted as information availability, not semantic drift. The categorical results support session-level measurement for AI search. They do not estimate the causal effect of history on model answers.

Majority voting over multiple LLM attempts improves mathematical reasoning, but correlated errors limit the effective sample size. A natural fix is to assign different reasoning strategies to different voters. The approach, Diverse Prompt Mixer, is tested on the AIMO 3 competition: 3 models, 23+ experiments, 50 IMO-level problems, one H100 80 GB, 5-hour limit. Every prompt-level intervention fails. High-temperature sampling already decorrelates errors; weaker strategies reduce accuracy more than they reduce correlation. Across an 8-point capability gap at equal N=8 and every optimization tested, model capability dominates. The gap between the best majority-vote score (42/50) and pass@20 (~45.5) is selection loss, not prompt loss. A verifier-based selector could close it. Prompt engineering cannot.

We describe our system for LitTraceQA (GroundLM @ EMNLP 2026): given a research question, retrieve the relevant papers from a pool of 27,487, cite the page and the table or figure where the answer lives, and answer in a requested format. Our main finding is that evidence grounding is limited by retrieval, not by reading. The page selector put the annotator's page, which we call the gold page, in front of the model that locates evidence only about half the time (52.6% gold-page recall), while that model, given the page, cited the right one in 45 of the 48 locators it emitted (94%). When the page was missing it rarely said so: of 45 such cases it returned nothing 14 times, a wrong page 24 times, and a correct page 7 times, so the pipeline failed quietly almost twice as often as it failed visibly. Since the failure was that the right page was never shown, the fix is to stop choosing: each retrieved paper fits in the model's context, so we show it whole. Page ranking survives only as a fallback inside papers too long to fit, which no test-split paper was, and gold-page recall reaches 100% on the papers we can parse. Separately, questions that identify their target by position rather than content, such as "the first author of the 24th reference", are served by parsing rather than retrieval: we resolve the bibliography into an addressable list, which also supplies identifiers the evidence metric scores. The final system scores 0.762 paper F_1, 0.441 evidence F_1 and 0.920 multiple-choice accuracy on the held-out test split. Because the pipeline depends on a closed model without seed control, we release a harness that verifies the paper's central claims against committed artifacts.

Reasoning distillation has emerged as an efficient and powerful paradigm for enhancing the reasoning capabilities of large language models. However, reasoning distillation may inadvertently cause benchmark contamination, where evaluation data included in distillation datasets can inflate performance metrics of distilled models. In this work, we formally define the task of distillation data detection, which is uniquely challenging due to the partial availability of distillation data. Then, we propose a novel and effective method Token Probability Deviation (TBD), which leverages the probability patterns of the generated output tokens. Our method is motivated by the analysis that distilled models tend to generate near-deterministic tokens for seen questions, while producing more low-probability tokens for unseen questions. Our key idea behind TBD is to quantify how far the generated tokens' probabilities deviate from a high reference probability. In effect, our method achieves competitive detection performance by producing lower scores for seen questions than for unseen questions. Extensive experiments demonstrate the effectiveness of our method, achieving an AUC of 0.918 and a TPR@1% FPR of 0.470 on the S1 dataset.

Background: Deep learning models are typically trained using stochastic gradient descent or one of its variants. These methods update the weights using their gradient, estimated from a small fraction of the training data. It has been observed that when using large batch sizes there is a persistent degradation in generalization performance - known as the "generalization gap" phenomena. Identifying the origin of this gap and closing it had remained an open problem. Contributions: We examine the initial high learning rate training phase. We find that the weight distance from its initialization grows logarithmically with the number of weight updates. We therefore propose a "random walk on random landscape" statistical model which is known to exhibit similar "ultra-slow" diffusion behavior. Following this hypothesis we conducted experiments to show empirically that the "generalization gap" stems from the relatively small number of updates rather than the batch size, and can be completely eliminated by adapting the training regime used. We further investigate different techniques to train models in the large-batch regime and present a novel algorithm named "Ghost Batch Normalization" which enables significant decrease in the generalization gap without increasing the number of updates. To validate our findings we conduct several additional experiments on MNIST, CIFAR-10, CIFAR-100 and ImageNet. Finally, we reassess common practices and beliefs concerning training of deep models and suggest they may not be optimal to achieve good generalization.

Most existing multi-hop datasets are extractive answer datasets, where the answers to the questions can be extracted directly from the provided context. This often leads models to use heuristics or shortcuts instead of performing true multi-hop reasoning. In this paper, we propose a new multi-hop dataset, MoreHopQA, which shifts from extractive to generative answers. Our dataset is created by utilizing three existing multi-hop datasets: HotpotQA, 2WikiMultihopQA, and MuSiQue. Instead of relying solely on factual reasoning, we enhance the existing multi-hop questions by adding another layer of questioning that involves one, two, or all three of the following types of reasoning: commonsense, arithmetic, and symbolic. Our dataset is created through a semi-automated process, resulting in a dataset with 1,118 samples that have undergone human verification. We then use our dataset to evaluate five different large language models: Mistral 7B, Gemma 7B, Llama 3 (8B and 70B), and GPT-4. We also design various cases to analyze the reasoning steps in the question-answering process. Our results show that models perform well on initial multi-hop questions but struggle with our extended questions, indicating that our dataset is more challenging than previous ones. Our analysis of question decomposition reveals that although models can correctly answer questions, only a portion - 38.7% for GPT-4 and 33.4% for Llama3-70B - achieve perfect reasoning, where all corresponding sub-questions are answered correctly. Evaluation code and data are available at https://github.com/Alab-NII/morehopqa

Generating diverse responses from large language models (LLMs) is crucial for applications such as planning/search and synthetic data generation, where diversity provides distinct answers across generations. Prior approaches rely on increasing temperature to increase diversity. However, contrary to popular belief, we show not only does this approach produce lower quality individual generations as temperature increases, but it depends on model's next-token probabilities being similar to the true distribution of answers. We propose , an alternative approach that uses the language model itself to partition the space into strata. At inference, a random stratum is selected and a sample drawn from within the strata. To measure diversity, we introduce CoverageQA, a dataset of underspecified questions with multiple equally plausible answers, and assess diversity by measuring KL Divergence between the output distribution and uniform distribution over valid ground truth answers. As computing probability per response/solution for proprietary models is infeasible, we measure recall on ground truth solutions. Our evaluation show using SimpleStrat achieves higher recall by 0.05 compared to GPT-4o and 0.36 average reduction in KL Divergence compared to Llama 3.

Retrieval-Augmented Generation equips large language models with the capability to retrieve external knowledge, thereby mitigating hallucinations by incorporating information beyond the model's intrinsic abilities. However, most prior works have focused on invoking retrieval deterministically, which makes it unsuitable for tasks such as long-form question answering. Instead, dynamically performing retrieval by invoking it only when the underlying LLM lacks the required knowledge can be more efficient. In this context, we delve deeper into the question, "To Retrieve or Not to Retrieve?" by exploring multiple uncertainty detection methods. We evaluate these methods for the task of long-form question answering, employing dynamic retrieval, and present our comparisons. Our findings suggest that uncertainty detection metrics, such as Degree Matrix Jaccard and Eccentricity, can reduce the number of retrieval calls by almost half, with only a slight reduction in question-answering accuracy.

Grounding large language models (LLMs) in verifiable external sources is a well-established strategy for generating reliable answers. Retrieval-augmented generation (RAG) is one such approach, particularly effective for tasks like question answering: it retrieves passages that are semantically related to the question and then conditions the model on this evidence. However, multi-hop questions, such as "Which company among NVIDIA, Apple, and Google made the biggest profit in 2023?," challenge RAG because relevant facts are often distributed across multiple documents rather than co-occurring in one source, making it difficult for standard RAG to retrieve sufficient information. To address this, we propose a RAG pipeline that incorporates question decomposition: (i) an LLM decomposes the original query into sub-questions, (ii) passages are retrieved for each sub-question, and (iii) the merged candidate pool is reranked to improve the coverage and precision of the retrieved evidence. We show that question decomposition effectively assembles complementary documents, while reranking reduces noise and promotes the most relevant passages before answer generation. Although reranking itself is standard, we show that pairing an off-the-shelf cross-encoder reranker with LLM-driven question decomposition bridges the retrieval gap on multi-hop questions and provides a practical, drop-in enhancement, without any extra training or specialized indexing. We evaluate our approach on the MultiHop-RAG and HotpotQA, showing gains in retrieval (MRR@10: +36.7%) and answer accuracy (F1: +11.6%) over standard RAG baselines.

AI-driven medical history-taking is an important component in symptom checking, automated patient intake, triage, and other AI virtual care applications. As history-taking is extremely varied, machine learning models require a significant amount of data to train. To overcome this challenge, existing systems are developed using indirect data or expert knowledge. This leads to a training-inference gap as models are trained on different kinds of data than what they observe at inference time. In this work, we present a two-stage re-ranking approach that helps close the training-inference gap by re-ranking the first-stage question candidates using a dialogue-contextualized model. For this, we propose a new model, global re-ranker, which cross-encodes the dialogue with all questions simultaneously, and compare it with several existing neural baselines. We test both transformer and S4-based language model backbones. We find that relative to the expert system, the best performance is achieved by our proposed global re-ranker with a transformer backbone, resulting in a 30% higher normalized discount cumulative gain (nDCG) and a 77% higher mean average precision (mAP).

Academic survey writing, which distills vast literature into a coherent and insightful narrative, remains a labor-intensive and intellectually demanding task. While recent approaches, such as general DeepResearch agents and survey-specialized methods, can generate surveys automatically (a.k.a. LLM4Survey), their outputs often fall short of human standards and there lacks a rigorous, reader-aligned benchmark for thoroughly revealing their deficiencies. To fill the gap, we propose a fine-grained, quiz-driven evaluation framework SurveyBench, featuring (1) typical survey topics source from recent 11,343 arXiv papers and corresponding 4,947 high-quality surveys; (2) a multifaceted metric hierarchy that assesses the outline quality (e.g., coverage breadth, logical coherence), content quality (e.g., synthesis granularity, clarity of insights), and non-textual richness; and (3) a dual-mode evaluation protocol that includes content-based and quiz-based answerability tests, explicitly aligned with readers' informational needs. Results show SurveyBench effectively challenges existing LLM4Survey approaches (e.g., on average 21% lower than human in content-based evaluation).

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.

Question generation (QG) from a given context can enhance comprehension, engagement, assessment, and overall efficacy in learning or conversational environments. Despite recent advancements in QG, the challenge of enhancing or measuring the diversity of generated questions often remains unaddressed. In this paper, we introduce a multi-question generation model (mQG), which is capable of generating multiple, diverse, and answerable questions by focusing on context and questions. To validate the answerability of the generated questions, we employ a SQuAD2.0 fine-tuned question answering model, classifying the questions as answerable or not. We train and evaluate mQG on the FairytaleQA dataset, a well-structured QA dataset based on storybooks, with narrative questions. We further apply a zero-shot adaptation on the TellMeWhy and SQuAD1.1 datasets. mQG shows promising results across various evaluation metrics, among strong baselines.

Eliciting information to reduce uncertainty about latent group-level properties from surveys and other collective assessments requires allocating limited questioning effort under real costs and missing data. Although large language models enable adaptive, multi-turn interactions in natural language, most existing elicitation methods optimize what to ask with a fixed respondent pool, and do not adapt respondent selection or leverage population structure when responses are partial or incomplete. To address this gap, we study adaptive group elicitation, a multi-round setting where an agent adaptively selects both questions and respondents under explicit query and participation budgets. We propose a theoretically grounded framework that combines (i) an LLM-based expected information gain objective for scoring candidate questions with (ii) heterogeneous graph neural network propagation that aggregates observed responses and participant attributes to impute missing responses and guide per-round respondent selection. This closed-loop procedure queries a small, informative subset of individuals while inferring population-level responses via structured similarity. Across three real-world opinion datasets, our method consistently improves population-level response prediction under constrained budgets, including a >12% relative gain on CES at a 10% respondent budget.

We present a new dataset for Visual Question Answering (VQA) on document images called DocVQA. The dataset consists of 50,000 questions defined on 12,000+ document images. Detailed analysis of the dataset in comparison with similar datasets for VQA and reading comprehension is presented. We report several baseline results by adopting existing VQA and reading comprehension models. Although the existing models perform reasonably well on certain types of questions, there is large performance gap compared to human performance (94.36% accuracy). The models need to improve specifically on questions where understanding structure of the document is crucial. The dataset, code and leaderboard are available at docvqa.org

Evaluating Retrieval-Augmented Generation (RAG) in large language models (LLMs) is challenging because benchmarks can quickly become stale. Questions initially requiring retrieval may become answerable from pretraining knowledge as newer models incorporate more recent information during pretraining, making it difficult to distinguish evidence-based reasoning from recall. We introduce NeoQA (News Events for Out-of-training Question Answering), a benchmark designed to address this issue. To construct NeoQA, we generated timelines and knowledge bases of fictional news events and entities along with news articles and Q\&A pairs to prevent LLMs from leveraging pretraining knowledge, ensuring that no prior evidence exists in their training data. We propose our dataset as a new platform for evaluating evidence-based question answering, as it requires LLMs to generate responses exclusively from retrieved evidence and only when sufficient evidence is available. NeoQA enables controlled evaluation across various evidence scenarios, including cases with missing or misleading details. Our findings indicate that LLMs struggle to distinguish subtle mismatches between questions and evidence, and suffer from short-cut reasoning when key information required to answer a question is missing from the evidence, underscoring key limitations in evidence-based reasoning.

Recently, a large amount of work has focused on improving large language models' (LLMs') performance on reasoning benchmarks such as math and logic. However, past work has largely assumed that tasks are well-defined. In the real world, queries to LLMs are often underspecified, only solvable through acquiring missing information. We formalize this as a constraint satisfaction problem (CSP) with missing variable assignments. Using a special case of this formalism where only one necessary variable assignment is missing, we can rigorously evaluate an LLM's ability to identify the minimal necessary question to ask and quantify axes of difficulty levels for each problem. We present QuestBench, a set of underspecified reasoning tasks solvable by asking at most one question, which includes: (1) Logic-Q: Logical reasoning tasks with one missing proposition, (2) Planning-Q: PDDL planning problems with initial states that are partially-observed, (3) GSM-Q: Human-annotated grade school math problems with one missing variable assignment, and (4) GSME-Q: a version of GSM-Q where word problems are translated into equations by human annotators. The LLM is tasked with selecting the correct clarification question(s) from a list of options. While state-of-the-art models excel at GSM-Q and GSME-Q, their accuracy is only 40-50% on Logic-Q and Planning-Q. Analysis demonstrates that the ability to solve well-specified reasoning problems may not be sufficient for success on our benchmark: models have difficulty identifying the right question to ask, even when they can solve the fully specified version of the problem. Furthermore, in the Planning-Q domain, LLMs tend not to hedge, even when explicitly presented with the option to predict ``not sure.'' This highlights the need for deeper investigation into models' information acquisition capabilities.

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.

The integration of retrieved passages and large language models (LLMs), such as ChatGPTs, has significantly contributed to improving open-domain question answering. However, there is still a lack of exploration regarding the optimal approach for incorporating retrieved passages into the answer generation process. This paper aims to fill this gap by investigating different methods of combining retrieved passages with LLMs to enhance answer generation. We begin by examining the limitations of a commonly-used concatenation approach. Surprisingly, this approach often results in generating "unknown" outputs, even when the correct document is among the top-k retrieved passages. To address this issue, we explore four alternative strategies for integrating the retrieved passages with the LLMs. These strategies include two single-round methods that utilize chain-of-thought reasoning and two multi-round strategies that incorporate feedback loops. Through comprehensive analyses and experiments, we provide insightful observations on how to effectively leverage retrieved passages to enhance the answer generation capability of LLMs.

Retrieval augmented generation (RAG) with large language models (LLMs) for Question Answering (QA) entails furnishing relevant context within the prompt to facilitate the LLM in answer generation. During the generation, inaccuracies or hallucinations frequently occur due to two primary factors: inadequate or distracting context in the prompts, and the inability of LLMs to effectively reason through the facts. In this paper, we investigate whether providing aligned context via a carefully selected passage sequence leads to better answer generation by the LLM for multi-hop QA. We introduce, "GenSco", a novel approach of selecting passages based on the predicted decomposition of the multi-hop questions}. The framework consists of two distinct LLMs: (i) Generator LLM, which is used for question decomposition and final answer generation; (ii) an auxiliary open-sourced LLM, used as the scorer, to semantically guide the Generator for passage selection. The generator is invoked only once for the answer generation, resulting in a cost-effective and efficient approach. We evaluate on three broadly established multi-hop question answering datasets: 2WikiMultiHop, Adversarial HotPotQA and MuSiQue and achieve an absolute gain of 15.1 and 5.9 points in Exact Match score with respect to the best performing baselines over MuSiQue and 2WikiMultiHop respectively.

An isolated final user message is often treated as the query in evaluations of AI systems. In a conversation, however, the actionable request may be distributed across preceding turns. We directly test whether that omitted within-conversation context changes answers. For each of 180 English multi-turn conversations sampled from a governed commercial corpus and the public PRISM dataset, we hold the final user message and requested answer model constant while generating three answers: one from the full role-labelled conversation, one from the final message alone, and one from the final message plus a prefix-only reconstruction capped at 160 words. A separately requested judge model evaluates answers under randomized labels. The prespecified primary endpoint is a material difference that could change what the user does, rather than a difference in style or detail. After inverse-probability weighting to the eligible cohorts, the full-conversation and isolated-final answers differ materially in 44.7% of cases (95% bootstrap CI 33.8% to 56.1%). Full-conversation answers score 0.49 points higher on a 0 to 4 request-satisfaction scale (0.32 to 0.67). Adding the compressed prefix reduces the material-difference rate to 30.8% (20.2% to 42.1%), a 13.9-point reduction (4.9% to 24.1%), and reduces the mean satisfaction gap to 0.01 points (-0.12 to 0.13). Yet compression is not equivalent to the complete dialogue context: almost one third of answers remain materially different. An order-swapped repeat on 48 cases yields 91.7% agreement and kappa = 0.83 for the primary decision. The study concerns preceding turns in the same conversation and does not test persistent memory across separate conversations.

In this paper we study yes/no questions that are naturally occurring --- meaning that they are generated in unprompted and unconstrained settings. We build a reading comprehension dataset, BoolQ, of such questions, and show that they are unexpectedly challenging. They often query for complex, non-factoid information, and require difficult entailment-like inference to solve. We also explore the effectiveness of a range of transfer learning baselines. We find that transferring from entailment data is more effective than transferring from paraphrase or extractive QA data, and that it, surprisingly, continues to be very beneficial even when starting from massive pre-trained language models such as BERT. Our best method trains BERT on MultiNLI and then re-trains it on our train set. It achieves 80.4% accuracy compared to 90% accuracy of human annotators (and 62% majority-baseline), leaving a significant gap for future work.

Retrieval-augmented generation (RAG) grounds answers in retrieved passages, yet relevance does not guarantee sufficiency: a topical passage may still fail to justify the answer. We study evidence sufficiency verification for selective RAG answering, in which a verifier receives a question, a candidate answer, and retrieved evidence and decides whether the evidence supports, refutes, or is insufficient for the answer, answering only when support is established. We present SURE-RAG, an aggregation protocol that treats evidence sufficiency as a set-level property: missing hops and unresolved conflicts cannot be detected by scoring passages independently. A shared claim-evidence verifier produces a local relation distribution for each (claim, passage) pair, which SURE-RAG aggregates into four interpretable answer-level feature blocks (coverage, relation strength, uncertainty, and retrieval), producing a three-way decision and an auditable selective score. We evaluate on HotpotQA-RAG v3, a controlled multi-hop benchmark, under an artifact-aware protocol (shortcut baselines, counterfactual swaps, no-oracle checks, and GPT-4o audits). Calibrated SURE-RAG attains 0.9075 Macro-F1 (raw 0.8951 +/- 0.0069), well above DeBERTa mean-pooling (0.6516) and a GPT-4o judge (0.7284), and on par with a strong concat cross-encoder (0.8888 +/- 0.0109) while remaining fully auditable. At 30% coverage, risk falls from 0.2588 to 0.1642, a 37% relative reduction. As a boundary-mapping experiment, we contrast SURE-RAG with GPT-4o on HaluBench unsafe detection: the ranking reverses (0.3343 vs. 0.7389 unsafe-F1), indicating that controlled sufficiency verification and natural hallucination detection are distinct problems.

This article proposes a new approach for assessing the quality of answers in political question-and-answer sessions. We measure the quality of an answer based on how easily and accurately it can be recognized in a random set of candidate answers given the question's text. This measure reflects the answer's relevance and depth of engagement with the question. Like semantic search, we can implement this approach by training a language model on the corpus of observed questions and answers without additional human-labeled data. We showcase and validate our methodology within the context of the Question Period in the Canadian House of Commons. Our analysis reveals that while some answers have a weak semantic connection to questions, hinting at some evasion or obfuscation, they are generally at least moderately relevant, far exceeding what we would expect from random replies. We also find a meaningful correlation between answer quality and the party affiliation of the members of Parliament asking the questions.

As language models shift from single-shot answer generation toward multi-step reasoning that retrieves and consumes evidence mid-inference, evaluating the role of individual retrieved items becomes more important. Existing RAG evaluation typically targets final-answer quality, citation faithfulness, or answer-level attribution, but none of these directly targets the intervention-based, per-evidence-item utility view we study here. We introduce CUE-R, a lightweight intervention-based framework for measuring per-evidence-item operational utility in single-shot RAG using shallow observable retrieval-use traces. CUE-R perturbs individual evidence items via REMOVE, REPLACE, and DUPLICATE operators, then measures changes along three utility axes (correctness, proxy-based grounding faithfulness, and confidence error) plus a trace-divergence signal. We also outline an operational evidence-role taxonomy for interpreting intervention outcomes. Experiments on HotpotQA and 2WikiMultihopQA with Qwen-3 8B and GPT-5.2 reveal a consistent pattern: REMOVE and REPLACE substantially harm correctness and grounding while producing large trace shifts, whereas DUPLICATE is often answer-redundant yet not fully behaviorally neutral. A zero-retrieval control confirms that these effects arise from degradation of meaningful retrieval. A two-support ablation further shows that multi-hop evidence items can interact non-additively: removing both supports harms performance far more than either single removal. Our results suggest that answer-only evaluation misses important evidence effects and that intervention-based utility analysis is a practical complement for RAG evaluation.

A common way of assessing language learners' mastery of vocabulary is via multiple-choice cloze (i.e., fill-in-the-blank) questions. But the creation of test items can be laborious for individual teachers or in large-scale language programs. In this paper, we evaluate a new method for automatically generating these types of questions using large language models (LLM). The VocaTT (vocabulary teaching and training) engine is written in Python and comprises three basic steps: pre-processing target word lists, generating sentences and candidate word options using GPT, and finally selecting suitable word options. To test the efficiency of this system, 60 questions were generated targeting academic words. The generated items were reviewed by expert reviewers who judged the well-formedness of the sentences and word options, adding comments to items judged not well-formed. Results showed a 75% rate of well-formedness for sentences and 66.85% rate for suitable word options. This is a marked improvement over the generator used earlier in our research which did not take advantage of GPT's capabilities. Post-hoc qualitative analysis reveals several points for improvement in future work including cross-referencing part-of-speech tagging, better sentence validation, and improving GPT prompts.

We demonstrate that a neural network pre-trained on text and fine-tuned on code solves mathematics course problems, explains solutions, and generates new questions at a human level. We automatically synthesize programs using few-shot learning and OpenAI's Codex transformer and execute them to solve course problems at 81% automatic accuracy. We curate a new dataset of questions from MIT's largest mathematics courses (Single Variable and Multivariable Calculus, Differential Equations, Introduction to Probability and Statistics, Linear Algebra, and Mathematics for Computer Science) and Columbia University's Computational Linear Algebra. We solve questions from a MATH dataset (on Prealgebra, Algebra, Counting and Probability, Intermediate Algebra, Number Theory, and Precalculus), the latest benchmark of advanced mathematics problems designed to assess mathematical reasoning. We randomly sample questions and generate solutions with multiple modalities, including numbers, equations, and plots. The latest GPT-3 language model pre-trained on text automatically solves only 18.8% of these university questions using zero-shot learning and 30.8% using few-shot learning and the most recent chain of thought prompting. In contrast, program synthesis with few-shot learning using Codex fine-tuned on code generates programs that automatically solve 81% of these questions. Our approach improves the previous state-of-the-art automatic solution accuracy on the benchmark topics from 8.8% to 81.1%. We perform a survey to evaluate the quality and difficulty of generated questions. This work is the first to automatically solve university-level mathematics course questions at a human level and the first work to explain and generate university-level mathematics course questions at scale, a milestone for higher education.

Recent large language models often answer factual questions correctly. But users can't trust any given claim a model makes without fact-checking, because language models can hallucinate convincing nonsense. In this work we use reinforcement learning from human preferences (RLHP) to train "open-book" QA models that generate answers whilst also citing specific evidence for their claims, which aids in the appraisal of correctness. Supporting evidence is drawn from multiple documents found via a search engine, or from a single user-provided document. Our 280 billion parameter model, GopherCite, is able to produce answers with high quality supporting evidence and abstain from answering when unsure. We measure the performance of GopherCite by conducting human evaluation of answers to questions in a subset of the NaturalQuestions and ELI5 datasets. The model's response is found to be high-quality 80\% of the time on this Natural Questions subset, and 67\% of the time on the ELI5 subset. Abstaining from the third of questions for which it is most unsure improves performance to 90\% and 80\% respectively, approaching human baselines. However, analysis on the adversarial TruthfulQA dataset shows why citation is only one part of an overall strategy for safety and trustworthiness: not all claims supported by evidence are true.

Despite extensive research on a wide range of question answering (QA) systems, most existing work focuses on answer containment-i.e., assuming that answers can be directly extracted and/or generated from documents in the corpus. However, some questions require inference, i.e., deriving answers that are not explicitly stated but can be inferred from the available information. We introduce Inferential QA -- a new task that challenges models to infer answers from answer-supporting passages which provide only clues. To study this problem, we construct QUIT (QUestions requiring Inference from Texts) dataset, comprising 7,401 questions and 2.4M passages built from high-convergence human- and machine-authored hints, labeled across three relevance levels using LLM-based answerability and human verification. Through comprehensive evaluation of retrievers, rerankers, and LLM-based readers, we show that methods effective on traditional QA tasks struggle in inferential QA: retrievers underperform, rerankers offer limited gains, and fine-tuning provides inconsistent improvements. Even reasoning-oriented LLMs fail to outperform smaller general-purpose models. These findings reveal that current QA pipelines are not yet ready for inference-based reasoning. Inferential QA thus establishes a new class of QA tasks that move towards understanding and reasoning from indirect textual evidence.

Long-term conversational memory requires retrieving evidence scattered across multiple sessions, yet single-pass retrieval fails on temporal and multi-hop questions. Existing iterative methods refine queries via generated content or document-level signals, but none explicitly diagnoses the evidence gap, namely what is missing from the accumulated retrieval set, leaving query refinement untargeted. We present EviMem, combining IRIS (Iterative Retrieval via Insufficiency Signals), a closed-loop framework that detects evidence gaps through sufficiency evaluation, diagnoses what is missing, and drives targeted query refinement, with LaceMem (Layered Architecture for Conversational Evidence Memory), a coarse-to-fine memory hierarchy supporting fine-grained gap diagnosis. On LoCoMo, EviMem improves Judge Accuracy over MIRIX on temporal (73.3% to 81.6%) and multi-hop (65.9% to 85.2%) questions at 4.5x lower latency. Code: https://github.com/AIGeeksGroup/EviMem.

In current NLP research, large-scale language models and their abilities are widely being discussed. Some recent works have also found notable failures of these models. Often these failure examples involve complex reasoning abilities. This work focuses on a simple commonsense ability, reasoning about when an action (or its effect) is feasible. To this end, we introduce FeasibilityQA, a question-answering dataset involving binary classification (BCQ) and multi-choice multi-correct questions (MCQ) that test understanding of feasibility. We show that even state-of-the-art models such as GPT-3, GPT-2, and T5 struggle to answer the feasibility questions correctly. Specifically, on MCQ and BCQ questions, GPT-3 achieves an accuracy of just (19%, 62%) and (25%, 64%) in zero-shot and few-shot settings, respectively. We also evaluate models by providing relevant knowledge statements required to answer the question. We find that the additional knowledge leads to a 7% gain in performance, but the overall performance still remains low. These results make one wonder how much commonsense knowledge about action feasibility is encoded in state-of-the-art models and how well they can reason about it.

We introduce a multi-stage prompting approach (MSP) for the generation of multiple choice questions (MCQs), harnessing the capabilities of GPT models such as text-davinci-003 and GPT-4, renowned for their excellence across various NLP tasks. Our approach incorporates the innovative concept of chain-of-thought prompting, a progressive technique in which the GPT model is provided with a series of interconnected cues to guide the MCQ generation process. Automated evaluations consistently demonstrate the superiority of our proposed MSP method over the traditional single-stage prompting (SSP) baseline, resulting in the production of high-quality distractors. Furthermore, the one-shot MSP technique enhances automatic evaluation results, contributing to improved distractor generation in multiple languages, including English, German, Bengali, and Hindi. In human evaluations, questions generated using our approach exhibit superior levels of grammaticality, answerability, and difficulty, highlighting its efficacy in various languages.

Generative models have recently started to outperform extractive models in Open Domain Question Answering, largely by leveraging their decoder to attend over multiple encoded passages and combining their information. However, generative models tend to be larger than extractive models due to the need for a decoder, run slower during inference due to auto-regressive decoder beam search, and their generated output often suffers from hallucinations. We propose to extend transformer encoders with the ability to fuse information from multiple passages, using global representation to provide cross-sample attention over all tokens across samples. Furthermore, we propose an alternative answer span probability calculation to better aggregate answer scores in the global space of all samples. Using our proposed method, we outperform the current state-of-the-art method by 2.5 Exact Match score on the Natural Question dataset while using only 25% of parameters and 35% of the latency during inference, and 4.4 Exact Match on WebQuestions dataset. When coupled with synthetic data augmentation, we outperform larger models on the TriviaQA dataset as well. The latency and parameter savings of our method make it particularly attractive for open-domain question answering, as these models are often compute-intensive.

The recent explosion in the capabilities of large language models has led to a wave of interest in how best to prompt a model to perform a given task. While it may be tempting to simply choose a prompt based on average performance on a validation set, this can lead to a deployment where unexpectedly poor responses are generated, especially for the worst-off users. To mitigate this prospect, we propose Prompt Risk Control, a lightweight framework for selecting a prompt based on rigorous upper bounds on families of informative risk measures. We offer methods for producing bounds on a diverse set of metrics, including quantities that measure worst-case responses and disparities in generation quality across the population of users. In addition, we extend the underlying statistical bounding techniques to accommodate the possibility of distribution shifts in deployment. Experiments on applications such as open-ended chat, medical question summarization, and code generation highlight how such a framework can foster responsible deployment by reducing the risk of the worst outcomes.

Generative models for open domain question answering have proven to be competitive, without resorting to external knowledge. While promising, this approach requires to use models with billions of parameters, which are expensive to train and query. In this paper, we investigate how much these models can benefit from retrieving text passages, potentially containing evidence. We obtain state-of-the-art results on the Natural Questions and TriviaQA open benchmarks. Interestingly, we observe that the performance of this method significantly improves when increasing the number of retrieved passages. This is evidence that generative models are good at aggregating and combining evidence from multiple passages.

The evaluation of Large Language Models (LLMs) on mathematical reasoning has largely focused on elementary problems, competition-style questions, or formal theorem proving, leaving graduate-level and computational mathematics relatively underexplored. We introduce CompMath-MCQ, a new benchmark dataset for assessing LLMs on advanced mathematical reasoning in a multiple-choice setting. The dataset consists of 1{,}500 originally authored questions by professors of graduate-level courses, covering topics including Linear Algebra, Numerical Optimization, Vector Calculus, Probability, and Python-based scientific computing. Three option choices are provided for each question, with exactly one of them being correct. To ensure the absence of data leakage, all questions are newly created and not sourced from existing materials. The validity of questions is verified through a procedure based on cross-LLM disagreement, followed by manual expert review. By adopting a multiple-choice format, our dataset enables objective, reproducible, and bias-free evaluation through lm_eval library. Baseline results with state-of-the-art LLMs indicate that advanced computational mathematical reasoning remains a significant challenge. We release CompMath-MCQ at the following link: https://github.com/biancaraimondi/CompMath-MCQ.git

We present TriviaQA, a challenging reading comprehension dataset containing over 650K question-answer-evidence triples. TriviaQA includes 95K question-answer pairs authored by trivia enthusiasts and independently gathered evidence documents, six per question on average, that provide high quality distant supervision for answering the questions. We show that, in comparison to other recently introduced large-scale datasets, TriviaQA (1) has relatively complex, compositional questions, (2) has considerable syntactic and lexical variability between questions and corresponding answer-evidence sentences, and (3) requires more cross sentence reasoning to find answers. We also present two baseline algorithms: a feature-based classifier and a state-of-the-art neural network, that performs well on SQuAD reading comprehension. Neither approach comes close to human performance (23% and 40% vs. 80%), suggesting that TriviaQA is a challenging testbed that is worth significant future study. Data and code available at -- http://nlp.cs.washington.edu/triviaqa/

The rise of large language models (LLMs) has enabled us to seek answers to inherently debatable questions on LLM chatbots, necessitating a reliable way to evaluate their ability. However, traditional QA benchmarks assume fixed answers are inadequate for this purpose. To address this, we introduce DebateQA, a dataset of 2,941 debatable questions, each accompanied by multiple human-annotated partial answers that capture a variety of perspectives. We develop two metrics: Perspective Diversity, which evaluates the comprehensiveness of perspectives, and Dispute Awareness, which assesses if the LLM acknowledges the question's debatable nature. Experiments demonstrate that both metrics align with human preferences and are stable across different underlying models. Using DebateQA with two metrics, we assess 12 popular LLMs and retrieval-augmented generation methods. Our findings reveal that while LLMs generally excel at recognizing debatable issues, their ability to provide comprehensive answers encompassing diverse perspectives varies considerably.

There is a growing literature on reasoning by large language models (LLMs), but the discussion on the uncertainty in their responses is still lacking. Our aim is to assess the extent of confidence that LLMs have in their answers and how it correlates with accuracy. Confidence is measured (i) qualitatively in terms of persistence in keeping their answer when prompted to reconsider, and (ii) quantitatively in terms of self-reported confidence score. We investigate the performance of three LLMs -- GPT4o, GPT4-turbo and Mistral -- on two benchmark sets of questions on causal judgement and formal fallacies and a set of probability and statistical puzzles and paradoxes. Although the LLMs show significantly better performance than random guessing, there is a wide variability in their tendency to change their initial answers. There is a positive correlation between qualitative confidence and accuracy, but the overall accuracy for the second answer is often worse than for the first answer. There is a strong tendency to overstate the self-reported confidence score. Confidence is only partially explained by the underlying token-level probability. The material effects of prompting on qualitative confidence and the strong tendency for overconfidence indicate that current LLMs do not have any internally coherent sense of confidence.

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 present a novel method for obtaining high-quality, domain-targeted multiple choice questions from crowd workers. Generating these questions can be difficult without trading away originality, relevance or diversity in the answer options. Our method addresses these problems by leveraging a large corpus of domain-specific text and a small set of existing questions. It produces model suggestions for document selection and answer distractor choice which aid the human question generation process. With this method we have assembled SciQ, a dataset of 13.7K multiple choice science exam questions (Dataset available at http://allenai.org/data.html). We demonstrate that the method produces in-domain questions by providing an analysis of this new dataset and by showing that humans cannot distinguish the crowdsourced questions from original questions. When using SciQ as additional training data to existing questions, we observe accuracy improvements on real science exams.

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

Uncertainty Quantification (UQ) research has primarily focused on closed-book factual question answering (QA), while contextual QA remains unexplored, despite its importance in real-world applications. In this work, we focus on UQ for the contextual QA task and propose a theoretically grounded approach to quantify epistemic uncertainty. We begin by introducing a task-agnostic, token-level uncertainty measure defined as the cross-entropy between the predictive distribution of the given model and the unknown true distribution. By decomposing this measure, we isolate the epistemic component and approximate the true distribution by a perfectly prompted, idealized model. We then derive an upper bound for epistemic uncertainty and show that it can be interpreted as semantic feature gaps in the given model's hidden representations relative to the ideal model. We further apply this generic framework to the contextual QA task and hypothesize that three features approximate this gap: context-reliance (using the provided context rather than parametric knowledge), context comprehension (extracting relevant information from context), and honesty (avoiding intentional lies). Using a top-down interpretability approach, we extract these features by using only a small number of labeled samples and ensemble them to form a robust uncertainty score. Experiments on multiple QA benchmarks in both in-distribution and out-of-distribution settings show that our method substantially outperforms state-of-the-art unsupervised (sampling-free and sampling-based) and supervised UQ methods, achieving up to a 13-point PRR improvement while incurring a negligible inference overhead.

Multiple-choice question answering (MCQA) is often used to evaluate large language models (LLMs). To see if MCQA assesses LLMs as intended, we probe if LLMs can perform MCQA with choices-only prompts, where models must select the correct answer only from the choices. In three MCQA datasets and four LLMs, this prompt bests a majority baseline in 11/12 cases, with up to 0.33 accuracy gain. To help explain this behavior, we conduct an in-depth, black-box analysis on memorization, choice dynamics, and question inference. Our key findings are threefold. First, we find no evidence that the choices-only accuracy stems from memorization alone. Second, priors over individual choices do not fully explain choices-only accuracy, hinting that LLMs use the group dynamics of choices. Third, LLMs have some ability to infer a relevant question from choices, and surprisingly can sometimes even match the original question. We hope to motivate the use of stronger baselines in MCQA benchmarks, the design of robust MCQA datasets, and further efforts to explain LLM decision-making.

Recent progress in reasoning models suggests that generating plausible attempts for research-level mathematics may be within reach, but verification remains a bottleneck, consuming scarce expert time. We hypothesize that a meaningful solution should contain enough method-level information that, when applied to a neighborhood of related questions, it should yield better downstream performance than incorrect solutions. Building on this idea, we propose Consequence-Based Utility, an oracle-free evaluator that scores each candidate by testing its value as an in-context exemplar in solving related yet verifiable questions. Our approach is evaluated on an original set of research-level math problems, each paired with one expert-written solution and nine LLM-generated solutions. Notably, Consequence-Based Utility consistently outperforms reward models, generative reward models, and LLM judges on ranking quality. Specifically, for GPT-OSS-120B, it improves Acc@1 from 67.2 to 76.3 and AUC from 71.4 to 79.6, with similarly large AUC gains on GPT-OSS-20B (69.0 to 79.2). Furthermore, compared to LLM-Judges, it also exhibits a larger solver-evaluator gap, maintaining a stronger correct-wrong separation even on instances where the underlying solver often fails to solve.

Multiple choice question answering (MCQA) is popular for LLM evaluation due to its simplicity and human-like testing, but we argue for its reform. We first reveal flaws in MCQA's format, as it struggles to: 1) test generation/subjectivity; 2) match LLM use cases; and 3) fully test knowledge. We instead advocate for generative formats based on human testing-where LLMs construct and explain answers-better capturing user needs and knowledge while remaining easy to score. We then show even when MCQA is a useful format, its datasets suffer from: leakage; unanswerability; shortcuts; and saturation. In each issue, we give fixes from education, like rubrics to guide MCQ writing; scoring methods to bridle guessing; and Item Response Theory to build harder MCQs. Lastly, we discuss LLM errors in MCQA-robustness, biases, and unfaithful explanations-showing how our prior solutions better measure or address these issues. While we do not need to desert MCQA, we encourage more efforts in refining the task based on educational testing, advancing evaluations.

In an educational setting, an estimate of the difficulty of multiple-choice questions (MCQs), a commonly used strategy to assess learning progress, constitutes very useful information for both teachers and students. Since human assessment is costly from multiple points of view, automatic approaches to MCQ item difficulty estimation are investigated, yielding however mixed success until now. Our approach to this problem takes a different angle from previous work: asking various Large Language Models to tackle the questions included in two different MCQ datasets, we leverage model uncertainty to estimate item difficulty. By using both model uncertainty features as well as textual features in a Random Forest regressor, we show that uncertainty features contribute substantially to difficulty prediction, where difficulty is inversely proportional to the number of students who can correctly answer a question. In addition to showing the value of our approach, we also observe that our model achieves state-of-the-art results on the BEA publicly available dataset.

We present a large-scale dataset for the task of rewriting an ill-formed natural language question to a well-formed one. Our multi-domain question rewriting MQR dataset is constructed from human contributed Stack Exchange question edit histories. The dataset contains 427,719 question pairs which come from 303 domains. We provide human annotations for a subset of the dataset as a quality estimate. When moving from ill-formed to well-formed questions, the question quality improves by an average of 45 points across three aspects. We train sequence-to-sequence neural models on the constructed dataset and obtain an improvement of 13.2% in BLEU-4 over baseline methods built from other data resources. We release the MQR dataset to encourage research on the problem of question rewriting.

Knowledge gaps and hallucinations are persistent challenges for Large Language Models (LLMs), which generate unreliable responses when lacking the necessary information to fulfill user instructions. Existing approaches, such as Retrieval-Augmented Generation (RAG) and tool use, aim to address these issues by incorporating external knowledge. Yet, they rely on additional models or services, resulting in complex pipelines, potential error propagation, and often requiring the model to process a large number of tokens. In this paper, we present a scalable method that enables LLMs to access external knowledge without depending on retrievers or auxiliary models. Our approach uses constrained generation with a pre-built prefix-tree index. Triples from a Knowledge Graph are verbalized in textual facts, tokenized, and indexed in a prefix tree for efficient access. During inference, to acquire external knowledge, the LLM generates facts with constrained generation which allows only sequences of tokens that form an existing fact. We evaluate our proposal on Question Answering and show that it scales to large knowledge bases (800 million facts), adapts to domain-specific data, and achieves effective results. These gains come with minimal generation-time overhead. ReFactX code is available at https://github.com/rpo19/ReFactX.

Retrieval Augmented Generation (RAG) improves correctness of Question Answering (QA) and addresses hallucinations in Large Language Models (LLMs), yet greatly increase computational costs. Besides, RAG is not always needed as may introduce irrelevant information. Recent adaptive retrieval methods integrate LLMs' intrinsic knowledge with external information appealing to LLM self-knowledge, but they often neglect efficiency evaluations and comparisons with uncertainty estimation techniques. We bridge this gap by conducting a comprehensive analysis of 35 adaptive retrieval methods, including 8 recent approaches and 27 uncertainty estimation techniques, across 6 datasets using 10 metrics for QA performance, self-knowledge, and efficiency. Our findings show that uncertainty estimation techniques often outperform complex pipelines in terms of efficiency and self-knowledge, while maintaining comparable QA performance.

We explore uncertainty quantification in large language models (LLMs), with the goal to identify when uncertainty in responses given a query is large. We simultaneously consider both epistemic and aleatoric uncertainties, where the former comes from the lack of knowledge about the ground truth (such as about facts or the language), and the latter comes from irreducible randomness (such as multiple possible answers). In particular, we derive an information-theoretic metric that allows to reliably detect when only epistemic uncertainty is large, in which case the output of the model is unreliable. This condition can be computed based solely on the output of the model obtained simply by some special iterative prompting based on the previous responses. Such quantification, for instance, allows to detect hallucinations (cases when epistemic uncertainty is high) in both single- and multi-answer responses. This is in contrast to many standard uncertainty quantification strategies (such as thresholding the log-likelihood of a response) where hallucinations in the multi-answer case cannot be detected. We conduct a series of experiments which demonstrate the advantage of our formulation. Further, our investigations shed some light on how the probabilities assigned to a given output by an LLM can be amplified by iterative prompting, which might be of independent interest.

Multiple-choice questions (MCQ) are frequently used to assess large language models (LLMs). Typically, an LLM is given a question and selects the answer deemed most probable after adjustments for factors like length. Unfortunately, LLMs may inherently favor certain answer choice IDs, such as A/B/C/D, due to inherent biases of priori unbalanced probabilities, influencing the prediction of answers based on these IDs. Previous research has introduced methods to reduce this ''selection bias'' by simply permutating options on a few test samples and applying to new ones. Another problem of MCQ is the lottery ticket choice by ''random guessing''. The LLM does not learn particular knowledge, but the option is guessed correctly. This situation is especially serious for those small-scale LLMs. To address them, a more thorough approach involves shifting from MCQ to open-style questions, which can fundamentally eliminate selection bias and random guessing issues. However, transitioning causes its own set of challenges in (1) identifying suitable open-style questions and (2) validating the correctness of LLM open-style responses against human-annotated ground-truths. This work aims to tackle these significant difficulties, and establish a new LLM evaluation benchmark through entirely open-style questions. Consequently, we introduce the Open-LLM-Leaderboard to track various LLMs' performance and reflect true capability of them, such as GPT-4o/4/3.5, Claude 3, Gemini, etc. Our code and dataset are available at https://github.com/VILA-Lab/Open-LLM-Leaderboard.

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.

We investigate the mathematical capabilities of ChatGPT by testing it on publicly available datasets, as well as hand-crafted ones, and measuring its performance against other models trained on a mathematical corpus, such as Minerva. We also test whether ChatGPT can be a useful assistant to professional mathematicians by emulating various use cases that come up in the daily professional activities of mathematicians (question answering, theorem searching). In contrast to formal mathematics, where large databases of formal proofs are available (e.g., the Lean Mathematical Library), current datasets of natural-language mathematics, used to benchmark language models, only cover elementary mathematics. We address this issue by introducing a new dataset: GHOSTS. It is the first natural-language dataset made and curated by working researchers in mathematics that (1) aims to cover graduate-level mathematics and (2) provides a holistic overview of the mathematical capabilities of language models. We benchmark ChatGPT on GHOSTS and evaluate performance against fine-grained criteria. We make this new dataset publicly available to assist a community-driven comparison of ChatGPT with (future) large language models in terms of advanced mathematical comprehension. We conclude that contrary to many positive reports in the media (a potential case of selection bias), ChatGPT's mathematical abilities are significantly below those of an average mathematics graduate student. Our results show that ChatGPT often understands the question but fails to provide correct solutions. Hence, if your goal is to use it to pass a university exam, you would be better off copying from your average peer!

Self-consistency decoding enhances LLMs' performance on reasoning tasks by sampling diverse reasoning paths and selecting the most frequent answer. However, it is computationally expensive, as sampling many of these (lengthy) paths is required to increase the chances that the correct answer emerges as the most frequent one. To address this, we introduce Confidence-Informed Self-Consistency (CISC). CISC performs a weighted majority vote based on confidence scores obtained directly from the model. By prioritizing high-confidence paths, it can identify the correct answer with a significantly smaller sample size. When tested on nine models and four datasets, CISC outperforms self-consistency in nearly all configurations, reducing the required number of reasoning paths by over 40% on average. In addition, we introduce the notion of within-question confidence evaluation, after showing that standard evaluation methods are poor predictors of success in distinguishing correct and incorrect answers to the same question. In fact, the most calibrated confidence method proved to be the least effective for CISC. Lastly, beyond these practical implications, our results and analyses show that LLMs can effectively judge the correctness of their own outputs, contributing to the ongoing debate on this topic.

We present a new dataset for machine comprehension in the medical domain. Our dataset uses clinical case reports with around 100,000 gap-filling queries about these cases. We apply several baselines and state-of-the-art neural readers to the dataset, and observe a considerable gap in performance (20% F1) between the best human and machine readers. We analyze the skills required for successful answering and show how reader performance varies depending on the applicable skills. We find that inferences using domain knowledge and object tracking are the most frequently required skills, and that recognizing omitted information and spatio-temporal reasoning are the most difficult for the machines.

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.

In reasoning about sequential events it is natural to pose probabilistic queries such as "when will event A occur next" or "what is the probability of A occurring before B", with applications in areas such as user modeling, medicine, and finance. However, with machine learning shifting towards neural autoregressive models such as RNNs and transformers, probabilistic querying has been largely restricted to simple cases such as next-event prediction. This is in part due to the fact that future querying involves marginalization over large path spaces, which is not straightforward to do efficiently in such models. In this paper we introduce a general typology for predictive queries in neural autoregressive sequence models and show that such queries can be systematically represented by sets of elementary building blocks. We leverage this typology to develop new query estimation methods based on beam search, importance sampling, and hybrids. Across four large-scale sequence datasets from different application domains, as well as for the GPT-2 language model, we demonstrate the ability to make query answering tractable for arbitrary queries in exponentially-large predictive path-spaces, and find clear differences in cost-accuracy tradeoffs between search and sampling methods.

Large language models (LLMs) have demonstrated remarkable capabilities across various tasks but still face challenges such as hallucinations. One potential reason for hallucinations is the lack of relevant knowledge or context. Thus, a promising solution to mitigate this issue involves instructing LLMs to respond with "I do not know" when a question falls outside their knowledge domain or the provided context. However, in this work, we observed that LLMs struggle to admit their lack of knowledge, primarily due to existing instruction datasets designed to encourage specific answers. To improve large language models' capability to recognize the boundaries of their knowledge, we propose a novel approach called uncertainty-sensitive tuning. This method involves two-stage training designed for uncertainty recognition and prompt-sensitive activation. In the first stage, we guide the LLM to reject unknown questions. In the second stage, we recover the decreased performance in QA tasks by incorporating designed causal instructions. By leveraging this method, we aim to enhance the model's ability to identify areas of uncertainty. The experimental results demonstrate that our proposed uncertainty-sensitive tuning method significantly improves the performance of the Llama2-chat-7B model. Specifically, it achieves a substantial 34.7% improvement in handling questions involving knowledge gaps compared to the original model. Moreover, our approach outperforms GPT-4, exhibiting a 9.4% increase in overall performance. We open-source the model and code on GitHub.

Large Language Models (LLMs) have shown the impressive capability of answering questions in a wide range of scenarios. However, when LLMs face different types of questions, it is worth exploring whether LLMs are aware that some questions have limited answers and need to respond more deterministically but some do not. We refer to this as question awareness of LLMs. The lack of question awareness in LLMs leads to two phenomena that LLMs are: (1) too casual to answer non-open-ended questions or (2) too boring to answer open-ended questions. In this paper, we first evaluate the question awareness in LLMs. The experimental results show that LLMs have the issues of lacking awareness of questions in certain domains, e.g. factual knowledge, resulting in hallucinations during the generation. To mitigate these, we propose a method called Question Awareness Temperature Sampling (QuATS). This method enhances the question awareness of LLMs by adaptively adjusting the output distributions based on question features. The automatic adjustment in QuATS eliminates the need for manual temperature tuning in text generation and consistently improves model performance in various benchmarks.

Ideally Open-Domain Question Answering models should exhibit a number of competencies, ranging from simply memorizing questions seen at training time, to answering novel question formulations with answers seen during training, to generalizing to completely novel questions with novel answers. However, single aggregated test set scores do not show the full picture of what capabilities models truly have. In this work, we perform a detailed study of the test sets of three popular open-domain benchmark datasets with respect to these competencies. We find that 60-70% of test-time answers are also present somewhere in the training sets. We also find that 30% of test-set questions have a near-duplicate paraphrase in their corresponding training sets. Using these findings, we evaluate a variety of popular open-domain models to obtain greater insight into what extent they can actually generalize, and what drives their overall performance. We find that all models perform dramatically worse on questions that cannot be memorized from training sets, with a mean absolute performance difference of 63% between repeated and non-repeated data. Finally we show that simple nearest-neighbor models out-perform a BART closed-book QA model, further highlighting the role that training set memorization plays in these benchmarks