Context: Architectural Design Decisions (ADDs) capture the rationale behind the structure and evolution of software systems but are rarely documented explicitly, and are often hidden inside source code commits. Recovering them is important for Architectural Knowledge Management (AKM). Problem: Extracting ADDs from commits is challenging due to their implicit and unstructured nature. Large Language Models (LLMs) have shown strong capabilities in understanding code and text, yet their effectiveness for this task remains underexplored. Study: We present a preliminary study using four LLMs (Gemini 3 Pro, DeepSeek R1, Kimi K2, Qwen3) with zeroshot and fewshot prompting on 30 developer-written ADDs from open-source projects. We score outputs with ROUGE-L, BLEU, METEOR, and BERTScore, and one author manually reviews the Gemini outputs. Results: All models reach a BERT-F1 above 0.81, and fewshot prompting improves alignment (Gemini BERT-F1: 0.828 to 0.847). However, the generated ADDs are often too long, implementation-focused, and miss the rationale behind the decision. This highlights opportunities for architecture-aware LLM systems and automated AKM.
Mousumi Akter, Md. Faiyaz Abdullah Sayeedi, Nurul Labib Sayeedi +1cs.CL
Idiomatic expressions are an integral part of natural language, reflecting cultural nuances and posing unique challenges for computational models, particularly in low-resource languages. In this paper, we present the first large-scale benchmark dataset of Bangla idioms, complemented by a synthetic multiple-choice question (MCQ) dataset for idiom meaning identification. We conduct a comprehensive evaluation of recent large language models (LLMs) across three idiom-related tasks: paraphrasing, idiom span detection, and meaning identification, leveraging zero-shot and few-shot prompting strategies. Our results reveal substantial variability in model performance, with no single LLM consistently outperforming others across all tasks. Notably, Phi-4-mini-instruct excels in paraphrasing, Kimi-K2-32b-instruct in span detection, and Gemini-2.5-flash in meaning identification. We believe that our datasets and analyses will provide valuable resources to guide future research in improving LLM comprehension of idiomatic expressions, particularly in Bangla and other low-resource languages.
Large language models (LLMs) exhibit in-context learning capabilities, where they can learn new tasks from prompt contexts without weight updates. We compare the learning efficacies of two prominent modes of in-context learning: (1) learning from descriptions of rules (instruction following); and (2) learning from examples of input-output demonstrations (few-shot prompting). Through five learning tasks that cover diverse domains (games, arithmetic, linguistic inferences), we compare two modes of learning (rules vs. examples) specifying the same underlying task. We furthermore explore model and task properties that modulate the learning efficacies. We find that models generally learn more reliably from rules than from examples alone, and additional examples on top of rules or simply scaling up the number of examples do not lead to consistent and significant gains. Instruction tuning amplifies the benefit of rule-based learning while keeping example-based learning capacities intact. Surprisingly, we find no privileged effect of example-based learning in base models, and rules still lead to gains in algebraic task domains. Overall, the comparative efficacy of rules over examples is larger when the task recruits algebraic abstractions and computations, and smaller when the task requires distributional sensitivity and/or recruits parametric knowledge.
To address the sequential and evolving nature of time series, the Online Time Series Forecasting (OTSF) task has been extensively studied in multiple domains. Existing research focuses on adapting to non-stationary environments by employing memory buffer-based retrieval strategies. However, we observe that such frameworks struggle with long-term adaptation and fail to generalize to unseen patterns. To this end, we introduce CoSPOT, an LLM-based online time series forecasting framework that leverages a pre-trained LLM as the backbone online forecaster, motivated by its strong few-shot capabilities. For efficient online adaptation, CoSPOT keeps the LLM frozen and employs compositional spectral prompts grounded in frequency-domain bases to guide the model with the overall distribution of the input, thereby substantially reducing the number of parameters updated during the online phase. Specifically, CoSPOT decomposes time series into frequency bases and composes the corresponding spectral basis prompts according to their amplitudes, allowing unseen patterns to be represented as new combinations of learned basis prompts. Our extensive experiments on real-world datasets demonstrate the superiority and practicality of CoSPOT across challenging online scenarios, including extended online phases and cross-dataset settings with substantial distribution shifts. Our code is available at https://github.com/seungyoon-Choi/CoSPOT.
Large language models (LLMs) struggle to classify text into taxonomies with many semantically similar labels, as the distinctions are domain-specific and not captured by pre-training. To handle large label spaces, a common approach retrieves top-$K$ candidate labels by embedding similarity and prompt the LLM to choose among them. However, top-$K$ retrieval reduces the number of candidates but does not help the model tell similar ones apart. When two similar labels both appear as candidates, the model lacks the signal to choose correctly between them. We propose a framework that (1) identifies which label pairs the model struggles to distinguish, (2) expands the candidate set to include confusable labels, and (3) generates targeted rules to differentiate between similar candidates. The framework requires no fine-tuning, and the generated rules transfer to smaller, cheaper models. On three benchmarks (WOS, Flipkart, LEDGAR), our approach improves Macro F1 by up to 10.0pp over retrieval baselines, with smaller models (2B--20B) gaining up to 11.5pp via cross-model transfer.
Recent unified open-vocabulary detection (OVD) supports heterogeneous prompts, including text queries, visual exemplars, and their combinations, but often rely on increasingly complex designs such as heavy cross-modal fusion, staged training, and iterative annotation pipelines. We revisit whether such complexity is necessary in the era of stronger foundation models. Our finding is that unified OVD can be made substantially simpler with semantic-rich visual representations and scalable grounding supervision. We present OPUS (\textbf{O}pen-vocabulary, \textbf{P}rompt-\textbf{U}nified, \textbf{S}imple), a unified detector supporting text, interactive visual, generic visual, and mixed prompting within one framework. OPUS adopts a simple three-part design. Its model architecture combines a semantic-rich visual encoder, built on a DINOv3-ConvNeXt-B backbone with efficient hybrid encoding, with a prompt-aware decoder that avoids prompt-specific branches for unified prompt reasoning. OPUS is trained with a one-stage text-visual training strategy with Instance-level Contrastive Alignment (ICA), and is supported by a SAM3-based single-pass data engine for heterogeneous grounding supervision. Experiments on COCO, LVIS-minival, and ODinW35 show that OPUS achieves state-of-the-art Visual-I performance, reaching 68.1/69.2/54.7 AP, while maintaining balanced Text and Visual-G accuracy. OPUS also turns mixed prompting from interference into complementarity, improving over text or visual prompt alone. These results show that simplicity and strong unified prompting capability can be achieved together.
Credit-default prediction is an important task in financial decision making. Traditional methods use fitted classifiers such as logistic regression and random forests on tabular features. Large language models (LLMs) have recently been applied to this task through prompting. In this work we study how a fitted classifier and an LLM can be combined for credit-default prediction. We distinguish telling the LLM to imitate a classifier from using the classifier to build the prompt. We hypothesize that a fitted classifier can supply the ranking ability that an LLM prompt lacks. We experiment on the Default of Credit Card Clients dataset, and report recall, F1, and the area under the ROC and precision-recall curves, with bootstrap confidence intervals. We observe that a few-shot LLM has the highest recall (0.47) and F1 (0.50) of any single model but ranks worse than a random forest (AUC-ROC 0.72 against 0.79). Instructing the LLM to imitate a classifier gives no significant change. Pruning the prompt to the classifier's eight most important features raises recall by 0.071 and F1 by 0.032. Adding the classifier's predicted probability to the prompt raises the LLM's AUC-ROC from 0.72 to 0.78, matching the random forest, while keeping 0.118 higher recall than it. The reverse composition, and the use of several classifiers, do not help. We thus recommend a simple classifier-guided prompt for LLM-based credit prediction.
Mehrzad Tareh, Horacio Saggion, Stefan Bottcs.CL cs.AI
Text simplification aims to make complex texts easier to understand while preserving their original meaning. Recent large language models can perform simplification through prompting, but it remains unclear whether adding explicit linguistic information about sentence complexity to the prompt improves their outputs. We investigate this question for multilingual sentence-level Easy-to-Read simplification in Catalan, Spanish, and Italian. Using Qwen3-8B, we compare a baseline prompt, a gold-cue prompt enriched with gold linguistic cues, and a predicted-cue prompt enriched with automatically predicted cues. We evaluate the outputs using SARI, BLEU, chrF, and BERTScore, and complement this evaluation with a manual qualitative analysis. Predicted-cue prompting obtains the best overall scores across all four metrics, although the gains over the baseline are small. Gold-cue prompting does not consistently improve over the baseline, and results vary across languages. These findings indicate that cue-based prompting can influence multilingual Easy-to-Read simplification, but its benefits are modest, metric-dependent, and language-dependent.
Metaphor-identification performance can change markedly across datasets that differ in text distribution and annotation policy. We examine whether a fixed expert-informed procedure produces a more even cross-dataset profile than task-specific parameter adaptation. Four prespecified conditions are compared for Chinese sentence-level metaphor identification: BERT fine-tuning (BERT-FT), QLoRA-based large language model fine-tuning (LLM-FT), direct zero-shot LLM prompting (LLM-ZS), and zero-shot prompting with a frozen procedural Skill (Skill-ZS). The Skill operationalizes established criteria involving contextual meaning, basic meaning, contrast, and comparison. Evaluation covers CMRE Test and two external datasets, CCIME and CMC. Fine-tuned scores are means over three seeds, whereas each zero-shot score comes from one deterministic configuration. Fine-tuning remains strongest on the native test set: BERT-FT reaches 91.76 Macro-F1. LLM-FT has the highest external mean (83.52), while Skill-ZS is close at 82.92 and has both the highest external floor (82.64) and the smallest observed range across all three datasets (4.08 points). In the matched zero-shot comparison, adding the Skill reduces metaphorical predictions on every dataset. This sharply lowers false positives on CCIME but increases false negatives on CMRE Test and CMC. The results position expert-informed Skill prompting as a complementary route to more even observed cross-dataset performance, while fine-tuning retains its advantage in native-data accuracy. To our knowledge, this is the first study to compare an expert-informed procedural Skill with task-specific fine-tuning in the same cross-dataset evaluation of Chinese sentence-level metaphor identification.
The imperfective paradox provides a useful test of compositional semantic analysis. Recent work constructs an NLI benchmark and reports that models frequently infer completed telic events from progressive descriptions, attributing this behavior to a Teleological Bias. It further argues that prompting interventions cause a Calibration Crisis. We reexamine the benchmark and conclusions and show that it is substantially affected by conceptual and evaluation mis-specifications. We identify three conceptual mis-specifications. In particular, Aspectual Reduction affects the benchmark construction, analysis, experiments, and conclusions. Under a strict NLI standard, 76% of Group A instances do not explicitly rule out culmination. In our native-speaker annotation, 38% of Group A examples and 29% of the Group C examples were judged to permit an alternative interpretation. To control these issues and lexical variation, we construct Lexically Matched Minimal Pairs. At the evaluation level, we formulate event-semantic NLI as a Multi-step Reasoning Problem and assess both intermediate semantic decisions and final predictions. Our results show that models often do not affirm culmination but nevertheless accept the corresponding simple-past hypothesis, a pattern we characterize as Sufficiency Bias. We further show that prompting interventions produce a Decision Shift among labels without reliably improving the underlying semantic understanding and reasoning. Intermediate and oracle-guided analyses identify two additional failure modes: errors in compositional aspectual classification and Surface-form Attraction toward surface-associated answers. Our experiments on Qwen-7B with suitable prompts, GPT-5.4, and Qwen-72B provide initial evidence for the context sensitivity of aspectual classification and suggest that these models can achieve performance comparable to that of human annotators.
Real-world situation appearances can deviate from their underlying physical states, challenging the reliability of multimodal large language models (MLLMs) in practical applications. In this paper, we term this phenomenon situational illusions and investigate: (1) how MLLMs perform under such illusions, and (2) how to mitigate the limitations. We first develop a comprehensive where-what-how taxonomy that characterizes where situational illusions occur, what targets they take, and how they arise. Building on this taxonomy, we introduce MSIBench, a benchmark designed to assess the discrimination, understanding, and reasoning capabilities of MLLMs under situational illusions. Evaluations of 27 model configurations reveal that current MLLMs are highly vulnerable to these illusions and exhibit 6 typical failure modes related to visual observation, grounding, and reasoning. To mitigate the limitations, we build on the core idea of systematically inspecting and reasoning over visual evidence for contextual understanding, developing prompting for closed-source models and supervised fine-tuning for open-source models, respectively. These two simple yet effective methods improve model performances by 20% at most, suggesting a practical path toward more reliable multimodal perception and reasoning in complex real-world environments.
Afonso Baldo, Hugo Pitorro, Areti Vassilopoulos +5cs.CL
Users increasingly turn to large language models for emotional support, yet little is known about how these models actually conduct a psychotherapy interaction. We introduce an ontology of ten therapeutic moves: compact, function-based categories grounded in the MULTI-60 inventory, validated through an annotation campaign with five licensed psychologists, and scaled with a judge-based approach that matches expert agreement. Applying it to real counseling transcripts and model-led sessions, we compare the move distributions between human clinicians and a panel of frontier models. Models over-use inquiry at up to three times the human rate, neglect psychoeducation, and are strongly context-anchored: they carry forward strategies initiated by a human clinician but rarely initiate them themselves. Exposing the ontology as a set of tools roughly halves the mean deviation from the human move distribution and improves turn-level alignment with human therapist by 7-9 percentage points, without any fine-tuning.
Amogh Raina, Ilias Chalkidis, Daniel Hershcovich +1cs.CL cs.AI
Reasoning has become a standard technique and feature for contemporary LLMs; however, its application and quality in the context of demanding legal-oriented tasks, such as legal case forecasting, remain under explored. We investigate how LLMs reason in the context of legal case forecasting, using legal cases from the European Court of Human Rights (ECtHR) as a testbed. We evaluate OpenAI GPT 5.4, a recent top-tier LLM, by exploring alternative prompting strategies that are more or less suggestive of what counts as legally meaningful reasoning in the context of ECtHR jurisprudence. We present our findings derived from assessing the model's responses with both human and LLM evaluation. We find that the examined model scores far from ideal in legal reasoning, the model produces structurally complete but substantively shallow analyses, and that LLM-as-a-Judge evaluators are internally consistent yet align only weakly with our trained annotators, i.e., reliable but not a valid substitute for human evaluation. Overall, the expert-curated prompt leads to more comprehensive reasoning, which does not result in more accurate predictions compared to the other examined settings. Based on our findings, we urge the community not to rely solely on automated LLM-based evaluation and to avoid using task accuracy as an appropriate proxy for reasoning quality.
Clinical-AI guidance increasingly recommends prompting language models to reason with attention to diversity, equity, and inclusion (DEI). We measure a side effect that misrepresents patients: a one-sentence DEI prompt appended to a medical question leads models to add patient demographic attributes (race, socioeconomic status, sex) the question never stated, in effect rewriting who the patient is. We call this demographic injection. Across 47 models, four medical benchmarks, and 376,000 responses scored by a validated model-judge pipeline, a single DEI prompt raises the injection rate from 0.7% to 33.1% (47x) in all 47 of 47 models, attributable to the equity content rather than to added length (18x above a length-matched control; p=1.4x10^-14). Most added content is a general population statement that leaves the answer unchanged, but a smaller subset attaches an attribute to the specific patient or changes the selected option (0.25-2.4% of responses, 99.8% toward the incorrect option), where the invented demographic changes the answer the model recommends. Phrasing scales the effect from 14% to 56%. DEI prompts are just one example of a more general mechanism. Any instruction that nudges how a model reasons can make it add unrequested details, including details about the patient. Flagged outputs are treated as model errors under study, not clinical guidance.
Professional communication is increasingly mediated by LLMs - but do these models serve all users equally? We show that when prompts contain linguistic features more commonly used by women (hedges, tag questions, collective reference), they systematically elicit shorter, less sophisticated, and less formal responses across three document types and four models. These effects persist after controlling for prompt complexity and feature carry-over. Explicit gender cues like sign-off names are encoded in the same representational space as linguistic dialect - suggesting shared underlying mechanisms - yet linguistic register is far more influential, producing large, consistent effects where names produce none. Our results further reveal that post-hoc mitigation is challenging: because these patterns are culturally embedded and outside conscious control, users cannot easily avoid them through strategic self-presentation, and mechanistic analysis reveals that linguistic features are encoded in early transformer layers and entangled with other features. Our work calls for upstream consideration of the influences of linguistic variation to mitigate disparate impacts of LLM-mediated workplace communication.
Anik Pramanik, Murat Kantarcioglu, Vincent Oria +1cs.DB cs.AI cs.CL cs.IR
Prompting-based (\textit{i}.\textit{e}., non-fine-tuning) Text-to-SQL methods, where underlying large language model parameters are not changed for the task, face three problems: (\textit{i})~relying on coarse-grained schema information that may not reveal the fine-grained relationships needed to distinguish ambiguous columns, (\textit{ii})~not capturing recurring SQL-generation failures, and (\textit{iii})~suffering from omission, hallucination, or misplacement of conditions in complex questions. This paper develops \textsc{DexterSQL}, a prompting/non-fine-tuning-based Text-to-SQL system that improves SQL generation with three novel components: (\textit{i})~\emph{deep schema explorator} that identifies ambiguous columns, analyzes their individual and joint data distributions to uncover their relationships and the distinct role of each, (\textit{ii})~\emph{database-agnostic rule creator} that mines mismatches between generated and gold SQL only on the training database and converts them into database-agnostic corrective rules that capture recurring LLM failure patterns; and (\textit{iii})~\emph{multi-path SQL generation} that introduces a dependency-tree-based intermediate representation that uses the question's sentence structure to guide its decomposition into an SQL skeleton for final SQL generation. \textsc{DexterSQL} achieves a higher accuracy compared to the state-of-the-art using both open-source/weight and closed-source/weight models. Particularly, \textsc{DexterSQL}'s shows a high improvement of at least 2.7\% using an open-weight model (GPT-OSS-120B) on BIRD-Dev, with total accuracy 67.6\%. \textsc{DexterSQL} also shows better improvement of at least 0.9\% using closed-weight models, with total accuracy 71.6\% and 72.2\% on BIRD-Dev with GPT-4o and GPT-5.2.
Roni Blushtein-Livnon, Tal Svoray, Osher Rafaeli +4cs.CV
Spatio-temporal PV data are essential for understanding adoption processes in off-grid regions, yet such data remain largely unavailable. Automated segmentation of remote sensing (RS) imagery offers a promising solution; yet, residential PV systems remain challenging targets because of their small size and sparse distribution, resulting in severe target-background imbalance. Vision-language foundation models (FMs) provide a data-efficient paradigm through prompt-based semantic and spatial guidance, but the relative contribution of different prompt types remains unclear. We systematically evaluate SAM3 for small-scale PV segmentation in RS imagery by comparing textual, geometric, and hybrid prompting, under varying supervision levels, training strategies, spatial resolutions, and imaging conditions. Multi-temporal aerial imagery from a large off-grid rural region serves as a study site, with findings validated across three additional datasets. Prompting strategy emerged as the dominant factor governing model behavior. Textual prompting consistently produced the lowest performance and showed the greatest sensitivity to supervision and imaging conditions. In contrast, spatial guidance substantially improved both segmentation accuracy and robustness. Hybrid prompting achieved the highest accuracy and stability, indicating that semantic and spatial guidance provide complementary information. Most performance gains were achieved with only a few hundred annotated samples, demonstrating strong data efficiency. Transfer learning had limited overall impact, with only modest improvements observed for textual prompting under limited supervision. Overall, our findings establish prompting strategy as a key determinant of SAM3 adaptation, robustness, and generalization, highlighting the potential of promptable FMs for scalable PV mapping in data-constrained off-grid regions.
How expressive is prompting a transformer? Answering this question is important for separating the roles of prompting, architecture, and pretraining in transformer models, and for determining whether task-specific behavior must be stored in model weights or can instead be induced at inference time through the prompt. We show, in an approximation-theoretic sense, that pretraining is optional: a single-layer softmax attention network with random, untrained weights can approximate any Hölder function on a compact manifold when steered by an appropriate soft prompt. Guided by the connection between softmax attention and kernel methods, we construct explicit soft prompts (a prompt per target function, independent of the query) as solutions to linear systems matching attention logits to Gaussian kernel exponents, under which the frozen transformer emulates the classical Nadaraya-Watson kernel estimator. The construction requires only a mild rank condition on the weights, which we show holds almost surely under Gaussian initialization. The prompted network inherits the theoretical guarantees of kernel regression, leading to universal approximation theorems with minimax-optimal rates that depend on the intrinsic dimension. We further quantify the cost of prompting, exposing a tradeoff between the norm of the constructed soft prompt tokens, prompt length, and hidden dimension. Numerical experiments corroborate the constructions and predicted rates.
Node ranking is a fundamental problem in graph information retrieval, measuring the relative importance of nodes and supporting a wide range of applications such as influence analysis, recommendation, and graph-based retrieval augmented generation. However, exact computation of graph-based ranking measures is often computationally prohibitive at scale. Existing GNN-based ranking methods provide scalable approximations, but they are typically tailored to individual ranking criteria and require retraining for each downstream task, which limits their transferability and efficiency. Recent graph pre-training approaches aim to enable knowledge transfer across tasks, yet their learning objectives are largely misaligned with node ranking, resulting in suboptimal adaptability to ranking-oriented applications. To address these limitations, we propose PreGress, the first ranking-native pre-training and prompting framework for supporting a wide range of node ranking tasks. PreGress performs multi-task pre-training using our carefully designed objectives, including degree centrality prediction and attribute reconstruction, to jointly capture structural and attribute information. To support heterogeneous ranking criteria, we design lightweight, task-specific prompt modules that adapt a frozen ranking backbone to downstream tasks without full retraining. Experiments on six public graphs and two real-world query-to-item benchmarks---Yelp2018 and MovieLens-100K---together with a controlled five-criterion graph-access study demonstrate strong ranking quality with low task-specific state overhead.
Hakeem Hannoon, Andrew Zhao, Mihir Narayan +2cs.AI cs.CL
Conversational assistants increasingly rely on persistent long-term memory to personalize responses across sessions. However, when stored user information is reintroduced into the model context, it can also influence responses in inappropriate or unrelated settings. We study two such failure modes in memory-augmented LLMs: cross-domain leakage, where memories from one life domain affect responses in another, and memory-induced sycophancy, where stored user beliefs make models more likely to agree with the user rather than respond truthfully. We apply a simple inference-time modification to how memories are presented to the model, without changing the model or the memory contents. Across seven models on PersistBench, we compare the commonly used all-in context format, where memories are injected as an unstructured list, with structured formats that partition memories by domain. This simple modification consistently reduces cross-domain leakage while preserving utility, with our strongest method reducing leakage by $8.8\%$ on average relative to the baseline.
Task knowledge is commonly stored either as text in the prompt or as an update to model weights. Text is modular but must be interpreted on every use, while weight adaptation makes the resulting capability difficult to load, remove, or share independently. We introduce KV-Skill, a design space of external factorized operators that a frozen language model reads through a lightweight interface. KV-Skill supports two complementary paths. Registration converts an authored text skill into a text-derived operator and trains a shared per-backbone interface. Reward learning develops a compact latent operator directly from task outcomes, with or without an authored skill. Neither path adds positions to the prompt. Across ten benchmarks and four backbones from three model families, converting text to a KV-Skill consistently makes the same procedural knowledge more effective. On Qwen3.5-4B LiveMath, registration reaches 77.2 accuracy, compared with 23.4 for the source text skill, 52.0 for SkillOpt, and 64.5 for SoftSkill. Under matched reward training and parameter budgets, KV-Skill gives the best result in seven of eight matched settings against soft prefixes, prefix tuning, and LoRA. A post-hoc rank analysis further shows that text-derived operators retain nearly all of their benefit with one task-aligned direction per injection layer, while matched random directions fail. Finally, one shared interface retains three independently loadable KV-Skills without measurable forgetting. These results show that task knowledge can be acquired from text or experience, compressed into an external operator, and deployed separately from the backbone. Code is available at: https://github.com/shawnzhg/KV-Skill
LLM judges are often asked to extract criteria and evidence before choosing between candidate answers. This workflow assumes that the intermediate record preserves the information needed for a later verdict. For reasoning-capable models, visible field order does not reveal internal decision order, so we test an observable alternative: persist the evidence in one call and make it the exclusive input to the next. Across 24,000 judgments over HelpSteer3, FeedbackQA, and CoVal, we compare standard pairwise judging, structured one-call judging, two-call evidence locking, and three-call pointwise locking with Claude Sonnet 4.5 and GPT-5. Evidence locking reduces agreement with released human preferences by 4 to 6 percentage points and increases answer-order inconsistency by 8 to 10 points relative to structured one-call judging. Pointwise locking is also harmful, while structured evidence elicitation remains close to standard judging. The result holds for both judges and all three datasets. Persisted evidence can support auditability, but it should not replace the source answers at decision time.
Large language models can derive a plausible mathematical object yet still violate explicit requirements--for example, by omitting a modular reduction, returning a non-integer, or using the wrong encoded answer form. We introduce Constraint-First Reasoning (CFR), a training-free two-stage prompting protocol: Stage 1 extracts and summarizes constraints entailed by the problem, and Stage 2 solves while checking intermediate and final results against that summary. Routed-CFR activates the two-stage protocol only when a text-only regex router detects restrictive cues; otherwise it uses direct chain-of-thought (CoT). Across AIME, CMIMC, BRUMO, and AIMO_AMC, the method improves direct CoT on multiple backbones. We further report convention-controlled routing experiments, matched prompting baselines, problem-level paired tests, decoding robustness, constraint-quality audits, total-token accounting, and an OlympiadBench evaluation. These analyses position CFR as a targeted test-time intervention whose benefit depends on recoverable constraints and reliable Stage 1 extraction, rather than as a general-purpose replacement for mathematical reasoning.
Despite its cultural relevance and diffusion, queer slang remains underrepresented in Natural Language Processing research. Towards addressing this gap, we introduce Slang-Q, a manually curated dataset of naturally user-generated English sentences paired with queer slang terms and reference definitions, built upon a newly constructed taxonomy of 118 queer terms. We use this resource to conduct a first exploratory evaluation of language models on their ability to understand and define queer slang under varying prompting conditions. Slang-Q is intended as a basis for studying how current models handle sensitive, community-specific language and whether they can provide accurate and reliable information about such forms of identity and linguistic expression.
Denys Pushkin, Albert Q. Jiang, Aryo Lotfi +2cs.AI
Chain-of-Thought (CoT) prompting remains the standard baseline for evaluating models' reasoning abilities. Originally, this technique was introduced to elicit step-by-step reasoning from large language models (LLMs), which would otherwise tend to directly output the final answer. However, many modern LLMs produce CoT-style responses \textit{natively} when presented with reasoning tasks, which made us revisit the effectiveness of standard CoT prompting. We evaluate several modern mid-sized language models on a math problem-solving task and find that models specialized for reasoning achieve better performance in a simple zero-shot setting than when using few-shot CoT examples - significantly surpassing officially reported results at no additional cost (e.g., from $\sim$77\% to $\sim$84\% for Mathstral on GSM8K). For the tested general-purpose model, a zero-shot CoT prompt is also sufficient to outperform a few-shot CoT baseline. We attribute this to a `guidance-distraction' tradeoff: standard CoT prompting also demands style adaptation, formatting compliance, and potentially undesired contextualization, which can distract models from the core reasoning task. Our findings suggest that using standard CoT prompting increasingly acts as a source of distraction as models grow stronger.
Simulating human-like Theory of Mind (ToM) has been a longstanding problem in natural language processing (NLP). To address this, existing works introduce a reasoning step of event hiding (a.k.a. perspective-taking), where events unknown to a character are removed before question answering. However, resorting to event hiding for ToM reasoning presents a performance degradation issue due to the strict output format constraints involved in event hiding. To mitigate this issue, we propose generating perspective-taking outputs as free-form explanations without event hiding, but this poses a notable yet underexplored challenge: LLMs need to inhibit responses to events unknown to characters, because the absence of event hiding exposes LLMs to these events throughout reasoning. To address this challenge, we hypothesize and empirically verify that LLMs can achieve such inhibition if a character's lack of knowledge about events is made explicit during reasoning. Based on this finding, we introduce PICTURE, a new prompting method that enables LLMs to generate a character's lack of knowledge within free-form Chain-of-Thought (CoT). Experimental results show that PICTURE outperforms existing prompting methods by an average of 7.3% on false-belief tasks.
As large language models (LLMs) continue to advance in complex reasoning tasks, they have learned to heavily prioritize explicit conditions provided in the input. However, in everyday commonsense reasoning, this mechanism exposes a critical vulnerability which we term Salience Bias: models become easily hijacked by useless explicit distractors (e.g., numerical values), leading them to ignore the implicit physical or commonsense prerequisites of a task. A critical open question is whether this failure reflects a genuine gap in commonsense knowledge or merely its suppression under misleading task framing. To investigate this, we construct the SaliTrap Benchmark, a high-quality dataset across four trap dimensions. Evaluating 12 state-of-the-art LLMs, we find that all mainstream models suffer significantly from salience bias, with severity scaling with distractor density and detecting the trap often decoupled from actually avoiding it. Crucially, by re-eliciting the same models with the task framing stripped away, we show that this is overwhelmingly a failure of \textbf{knowledge suppression rather than knowledge absence}: a context-free knowledge probe alone recovers over 90\% of sycophantic-compliance failures, revealing that the requisite commonsense is intrinsically present but actively crowded out by salient distractors that lure the model into over-compliant, unnecessary computation. Building on this diagnosis, we further show that lightweight, inference-time prompting alone substantially closes the gap without any retraining. Our findings relocate the bottleneck of commonsense reasoning failures from model competence to elicitation, and we release SaliTrap as a testbed for this blind spot. The codes are available at https://github.com/Wuzheng02/SaliTrap.
We present dLLM-SetScore, a training-free method that uses discrete masked-diffusion language models for multi-label text classification. For each candidate label, it asks a short yes/no question and compares the probabilities of the two answer tokens at one masked position. The method uses no task-specific fine-tuning or training on textual-entailment datasets; a 200-example labelled validation slice selects thresholds, temperature, and prompt wording. We first show that placing all labels in one prompt creates a strong slot-position asymmetry: the first answer slot is predicted positive on $99.4\%$ of GoEmotions examples and $100\%$ of Reuters examples. Per-label scoring places every label in the same syntactic position, making predictions invariant to label order and avoiding this artifact. We evaluate LLaDA-8B and Dream-7B on six datasets against NLI models, an autoregressive LLM, SetFit, and supervised classifiers. On the five datasets shared by both diffusion families, Instruct checkpoints improve macro-F1 in 9 of 10 comparisons and micro-F1 in 8 of 10, although these comparisons do not identify the cause. Within our protocol, LLaDA-Instruct records the highest training-free values for both Reuters and ECtHR metrics. We prove permutation invariance, characterize thresholded decisions under weighted Hamming loss, and derive shortlist ceilings for recall and F1. An exploratory local Joint Set Refinement step lowers F1 from biased and unbiased initializations and is retained as a negative result.
Danyllo Albuquerque, José Renan, Guillermo Rodríguez +6cs.SE cs.AI
Microservice architectures have become dominant for modernizing monolithic systems, yet identifying appropriate services remains challenging and largely manual. Existing decomposition approaches are predominantly code-centric, limiting applicability in early design stages where only textual requirements are available. Despite advances in Large Language Models (LLMs), limited empirical evidence exists on their ability to synthesize complete microservice architectures from natural-language requirements, including service definitions and inter-service interactions. This study investigates whether an LLM can bridge requirements engineering and architectural design, generating architectures solely from textual requirements and evaluating structural agreement and perceived quality of results. We conduct a mixed-method study using OpenAI o3 under zero-shot (ZS) and few-shot (FS) prompting across two systems (Bookstore, PetClinic), one execution per system/condition. Architectures are evaluated through (i) comparison with reference architectures using precision, recall, and F1-score for service identification and communication recovery, and (ii) a blinded expert assessment of correctness, completeness, modularity, and plausibility, plus open feedback synthesis. OpenAI o3 identifies services with higher agreement under FS prompting (F1 = 0.79 for ZS versus = 0.97 for FS). Communication recovery is more challenging: ZS produces dense architectures with high recall but low precision (F1 = 0.61), while FS improves agreement, reaching F1 = 0.82 and reducing unsupported dependencies. Expert evaluation corroborates these results, with FS architectures perceived as more modular, coherent, and plausible than ZS outputs. OpenAI o3 shows potential for requirements-driven synthesis when guided by exemplar prompting. Results are model- and context-specific from two small systems, not model-independent proof.
Skill-based prompting has become a practical mechanism for improving large language model (LLM) agents, yet existing skill acquisition methods often treat skills as experience summaries, memory entries, or direct summaries of successful demonstrations. This creates a mismatch for weaker student agents: when a student fails because it lacks task knowledge or operational strategy, its failed trajectory may not contain enough evidence to infer the missing behavior, while the teacher trajectory may be too implicit to be internalized as reusable guidance. We propose SKILL-KD, a contrastive skill distillation framework that treats skills as an explicit distillation medium between agents of different capabilities. Given a student failure and the teacher trajectory on the same task, SKILL-KD distills their actionable discrepancy into a textual skill patch, evaluates the patch by re-running the student, and iteratively refines the patch when the student still fails. To prevent repeated local updates from causing skill drift, SKILL-KD further maintains trace-linked edit histories and performs Drift-Aware Skill Consolidation, deciding whether each patch should add a new rule, delete or modify an existing rule, or be skipped. Across five agent benchmarks and two student settings, SKILL-KD consistently improves frozen student agents over fixed-model adaptation baselines.