Large Language Models

Freight brokerages negotiate thousands of carrier rates daily under dynamic pricing conditions where models frequently revise targets mid-conversation. Classical time-dependent concession frameworks use a fixed shape parameter $β$ that cannot adapt to these updates. Deriving $β$ from the live spread enables adaptation but introduces a new problem: a pricing shift can cause the formula to retract a previous offer, violating monotonicity. LLM-powered brokers offer flexibility but require expensive reasoning models, produce non-deterministic pricing, and remain vulnerable to prompt injection. We propose a two-index anchor-and-resume framework that addresses both limitations. A spread-derived $β$ maps each load's margin structure to the correct concession posture, while the anchor-and-resume mechanism guarantees monotonically non-decreasing offers under arbitrary pricing shifts. All pricing decisions remain in a deterministic formula; the LLM, when used, serves only as a natural-language translation layer. Empirical evaluation across 115,125 negotiations shows that the adaptive $β$ tailors behavior by regime: in narrow spreads, it concedes quickly to prioritize deal closure and load coverage; in medium and wide spreads, it matches or exceeds the best fixed-$β$ baselines in broker savings. Against an unconstrained 20-billion-parameter LLM broker, it achieves similar agreement rates and savings. Against LLM-powered carriers as more realistic stochastic counterparties, it maintains comparable savings and higher agreement rates than against rule-based opponents. By decoupling the LLM from pricing logic, the framework scales horizontally to thousands of concurrent negotiations with negligible inference cost and transparent decision-making.

Large language models trained on human feedback may suppress fraud warnings when investors arrive already persuaded of a fraudulent opportunity. We tested this in a preregistered experiment across seven leading LLMs and twelve investment scenarios covering legitimate, high-risk, and objectively fraudulent opportunities, combining 3,360 AI advisory conversations with a 1,201-participant human benchmark. Contrary to predictions, motivated investor framing did not suppress AI fraud warnings; if anything, it marginally increased them. Endorsement reversal occurred in fewer than 3 in 1,000 observations. Human advisors endorsed fraudulent investments at baseline rates of 13-14%, versus 0% across all LLMs, and suppressed warnings under pressure at two to four times the AI rate. AI systems currently provide more consistent fraud warnings than lay humans in an identical advisory role.

While LLMs demonstrate impressive reasoning capabilities, they remain fragile in multi-step logical deduction, where a single transition error can propagate through the entire reasoning chain, leading to unstable performance. In this work, we identify logical connectives as primary points of this structural fragility. Through empirical analysis, we show that connective tokens function as high entropy forking points, at which models frequently struggle to determine the correct logical direction. Motivated by this observation, we hypothesize that intervening in logical connective selection can guide LLMs toward more correct logical direction, thereby improving the overall reasoning chain. To validate this hypothesis, we propose a multi-layered framework that intervenes specifically at these logic-critical junctions in the reasoning process. Our framework includes (1) Gradient-based Logical Steering to guide LLMs internal representations towards valid reasoning subspaces, (2) Localized Branching to resolve ambiguity via targeted look-ahead search, and (3) Targeted Transition Preference Optimization, a surgical reinforcement learning objective that selectively optimizes single-token preferences at logical pivots. Crucially, by concentrating intervention solely on logic-critical transitions, our framework achieves a favorable accuracy--efficiency trade-off compared to global inference time scaling methods like beam search and self-consistency.

Automatic Bloom's taxonomy classification of assessment questions can substantially reduce instructor workload, but labeling is subjective and teacher-dependent. Prior machine learning (ML) and deep learning (DL) approaches reported strong within-dataset results, yet were rarely evaluated in cross-dataset settings, leaving real-world generalizability unclear; meanwhile, LLM effectiveness for Bloom question classification has not been systematically studied. We evaluated the cross-dataset generalization of existing ML/DL methods and assessed LLMs with multiple prompting strategies on five datasets; the best prompting strategy combined in-context examples with course-specific action verbs. Supervised ML/DL models degraded substantially on unseen datasets, whereas LLMs were more stable, suggesting a robust alternative across diverse educational contexts. Based on the best prompting strategy, we also presented a lightweight UI that supports instructors in automatically classifying large question banks; a usability study indicated low workload and high usability.

Function calling empowers large language models (LLMs) to interface with external tools, yet existing RL-based approaches suffer from misalignment between reasoning processes and tool-call decisions. We propose R2IF, a reasoning-aware RL framework for interpretable function calling, adopting a composite reward integrating format/correctness constraints, Chain-of-Thought Effectiveness Reward (CER), and Specification-Modification-Value (SMV) reward, optimized via GRPO. Experiments on BFCL/ACEBench show R2IF outperforms baselines by up to 34.62% (Llama3.2-3B on BFCL) with positive Average CoT Effectiveness (0.05 for Llama3.2-3B), enhancing both function-calling accuracy and interpretability for reliable tool-augmented LLM deployment.

Automated feature generation extracts informative features from raw tabular data without manual intervention and is crucial for accurate, generalizable machine learning. Traditional methods rely on predefined operator libraries and cannot leverage task semantics, limiting their ability to produce diverse, high-value features for complex tasks. Recent Large Language Model (LLM)-based approaches introduce richer semantic signals, but still suffer from a restricted feature space due to fixed generation patterns and from the absence of feedback from the learning objective. To address these challenges, we propose a Memory-Augmented LLM-based Multi-Agent System (\textbf{MALMAS}) for automated feature generation. MALMAS decomposes the generation process into agents with distinct responsibilities, and a Router Agent activates an appropriate subset of agents per iteration, further broadening exploration of the feature space. We further integrate a memory module comprising procedural memory, feedback memory, and conceptual memory, enabling iterative refinement that adaptively guides subsequent feature generation and improves feature quality and diversity. Extensive experiments on multiple public datasets against state-of-the-art baselines demonstrate the effectiveness of our approach. The code is available at https://github.com/fxdong24/MALMAS

This paper proposes EvoAgent--an evolvable large language model (LLM) agent framework that integrates structured skill learning with a hierarchical sub-agent delegation mechanism. EvoAgent models skills as multi-file structured capability units equipped with triggering mechanisms and evolutionary metadata, and enables continuous skill generation and optimization through a user-feedback-driven closed-loop process. In addition, by incorporating a three-stage skill matching strategy and a three-layer memory architecture, the framework supports dynamic task decomposition for complex problems and long-term capability accumulation. Experimental results based on real-world foreign trade scenarios demonstrate that, after integrating EvoAgent, GPT5.2 achieves significant improvements in professionalism, accuracy, and practical utility. Under a five-dimensional LLM-as-Judge evaluation protocol, the overall average score increases by approximately 28%. Further model transfer experiments indicate that the performance of an agent system depends not only on the intrinsic capabilities of the underlying model, but also on the degree of synergy between the model and the agent architecture. Code, data, and documents will be released at https://github.com/Focus-AI-Center/Mentarc-EvoAgent.git.

Transformers suffer from a high computational cost that grows with sequence length for self-attention, making inference in long streams prohibited by memory consumption. Constant-memory alternatives such as RNNs and SSMs compress history into states with fixed size and thus lose long-tail dependencies, while methods that memorize contexts into parameters, such as Test-Time Training (TTT), are prone to overfitting token-level projection and fail to preserve the causal effect of context in pretrained LLMs. We propose Absorber LLM, which formulates long-context retention as a self-supervised causal synchronization: after absorbing historical contexts into parameters, a contextless model should match the original model with full context on future generations. We optimize this objective by synchronizing internal behaviors of the updated model with the original one, ensuring context absorption and generalization. Experiments on long-context and streaming benchmarks show that Absorber LLM reduces inference memory and improves accuracy over prior parameter-as-memory baselines.

We introduce IndiaFinBench, to our knowledge the first publicly available evaluation benchmark for assessing large language model (LLM) performance on Indian financial regulatory text. Existing financial NLP benchmarks draw exclusively from Western financial corpora (SEC filings, US earnings reports, English-language financial news), leaving a significant gap in coverage of non-Western regulatory frameworks. IndiaFinBench addresses this gap with 406 expert-annotated question-answer pairs drawn from 192 documents sourced from the Securities and Exchange Board of India (SEBI) and the Reserve Bank of India (RBI), spanning four task types: regulatory interpretation (174 items), numerical reasoning (92 items), contradiction detection (62 items), and temporal reasoning (78 items). Annotation quality is validated through a model-based secondary pass (kappa=0.918 on contradiction detection; 90.7% overall agreement on a 150-item subset) and a 180-item human inter-annotator agreement study across three annotation rounds (kappa=0.645 on contradiction detection; 77.2% overall agreement; 44.3% benchmark coverage). We evaluate twelve models under zero-shot conditions, with accuracy ranging from 70.4% (Gemma 4 E4B) to 89.7% (Gemini 2.5 Flash). All models substantially outperform a non-specialist human baseline of 69.0%. Numerical reasoning is the most discriminative task, with a 35.9 percentage-point spread across models. Bootstrap significance testing (10,000 resamples) reveals three statistically distinct performance tiers. The dataset, evaluation code, and all model outputs are available at https://github.com/rajveerpall/IndiaFinBench

Deploying large language models (LLMs) in resource-constrained environments is hindered by heavy computational and memory requirements. We present LBLLM, a lightweight binarization framework that achieves effective W(1+1)A4 quantization through a novel three-stage quantization strategy. The framework proceeds as follows: (1) initialize a high-quality quantized model via PTQ; (2) quantize binarized weights, group-wise bitmaps, and quantization parameters through layer-wise distillation while keeping activations in full precision; and (3) training learnable activation quantization factors to dynamically quantize activations to 4 bits. This decoupled design mitigates interference between weight and activation quantization, yielding greater training stability and better inference accuracy. LBLLM, trained only using 0.016B tokens with a single GPU, surpasses existing state-of-the-art binarization methods on W2A4 quantization settings across tasks of language modeling, commonsense QA, and language understanding. These results demonstrate that extreme low-bit quantization of LLMs can be both practical and highly effective without introducing any extra high-precision channels or rotational matrices commonly used in recent PTQ-based works, offering a promising path toward efficient LLM deployment in resource-limited situations.

Modern distributed systems generate massive volumes of log data that are critical for detecting anomalies and cyber threats. However, in real world settings, these logs are often distributed across multiple organizations and cannot be centralized due to privacy and security constraints. Existing log anomaly detection methods, including recent large language model (LLM) based approaches, largely rely on centralized training and are not suitable for such environments. In this paper, we propose DP-FLogTinyLLM, a privacy preserving federated framework for log anomaly detection using parameter efficient LLMs. Our approach enables collaborative learning without sharing raw log data by integrating federated optimization with differential privacy. To ensure scalability in resource constrained environments, we employ low rank adaptation (LoRA) for efficient fine tuning of Tiny LLMs at each client. Empirical results on the Thunderbird and BGL datasets show that the proposed framework matches the performance of centralized LLM based methods, while incurring additional computational overhead due to privacy mechanisms. Compared to existing federated baselines, DP-FLogTinyLLM consistently achieves higher precision and F1-score, with particularly strong gains on the Thunderbird dataset, highlighting its effectiveness in detecting anomalies while minimizing false positives.

We introduce Options LLM (OLLM), a simple, general method that replaces the single next-token prediction of standard LLMs with a \textit{set of learned options} for the next token, indexed by a discrete latent variable. Instead of relying on temperature or sampling heuristics to induce diversity, OLLM models variation explicitly: a small latent space parametrizes multiple plausible next-token options which can be selected or searched by a downstream policy. Architecturally, OLLM is a lightweight "plug-in" that inserts two layers: an encoder and a decoder, before the output head, allowing almost any pretrained LLM to be converted with minimal additional parameters. We apply OLLM to a 1.7B-parameter backbone (only $1.56\%$ of parameters trainable) trained on OpenMathReasoning and evaluated on OmniMath. The SOTA LoRA-adapted baselines peak at $51\%$ final answer correctness, while OLLM's option set allows up to $\sim 70\%$ under optimal latent selection. We then train a compact policy in the latent space that emits latents to control generation. Operating in a low-dimensional option space makes reward optimization far more sample-efficient and substantially reduces common misalignments (e.g., language switching or degenerate reasoning), as the policy is constrained to options learned during SFT. Crucially, this alignment arises from model structure rather than additional KL or handcrafted alignment losses. Our results demonstrate that optionized next-token modeling enhances controllability, robustness, and efficiency in math reasoning, and highlight latent-space policy learning as a promising direction for reinforcement learning in LLMs.

Statement autoformalization acts as a critical bridge between human mathematics and formal mathematics by translating natural language problems into formal language. While prior works have focused on data synthesis and diverse training paradigms to optimize end-to-end Large Language Models (LLMs), they typically treat formal code as flat sequences, neglecting the hierarchical logic inherent in mathematical statements. In this work, we introduce Decompose, Structure, and Repair (DSR), a neuro-symbolic framework that restructures autoformalization into a modular pipeline. DSR decomposes statements into logical components and maps them to structured operator trees, leveraging this topological blueprint to precisely localize and repair errors via sub-tree refinement. Furthermore, we introduce PRIME, a benchmark of 156 undergraduate and graduate-level theorems selected from canonical textbooks and expertly annotated in Lean 4. Experimental results demonstrate that DSR establishes a new state-of-the-art, consistently outperforming baselines under equivalent computational budgets. The datasets, model, and code are available at https://github.com/XiaoyangLiu-sjtu/DSR.

Knowledge graphs (KGs) are increasingly integrated with large language models (LLMs) to provide structured, verifiable reasoning. A core operation in this integration is multi-hop retrieval, yet existing systems struggle to balance efficiency, scalability, and interpretability. We introduce LogosKG, a novel, hardware-aligned framework that enables scalable and interpretable k-hop retrieval on large KGs by building on symbolic KG formulations and executing traversal as hardware-efficient operations over decomposed subject, object, and relation representations. To scale to billion-edge graphs, LogosKG integrates degree-aware partitioning, cross-graph routing, and on-demand caching. Experiments show substantial efficiency gains over CPU and GPU baselines without loss of retrieval fidelity. With proven performance in KG retrieval, a downstream two-round KG-LLM interaction demonstrates how LogosKG enables large-scale, evidence-grounded analysis of how KG topology, such as hop distribution and connectivity, shapes the alignment between structured biomedical knowledge and LLM diagnostic reasoning, thereby opening the door for next-generation KG-LLM integration. The source code is publicly available at https://github.com/LARK-NLP-Lab/LogosKG, and an online demo is available at https://lark-nlp-lab-logoskg.hf.space/.

Scientific feasibility assessment asks whether a claim is consistent with established knowledge and whether experimental evidence could support or refute it. We frame feasibility assessment as a diagnostic reasoning task in which, given a hypothesis, a model predicts feasible or infeasible and justifies its decision. We evaluate large language models (LLMs) under controlled knowledge conditions (hypothesis-only, with experiments, with outcomes, or both) and probe robustness by progressively removing portions of the experimental and/or outcome context. Across multiple LLMs and two datasets, providing outcome evidence is generally more reliable than providing experiment descriptions. Outcomes tend to improve accuracy beyond what internal knowledge alone provides, whereas experimental text can be brittle and may degrade performance when the context is incomplete. These findings clarify when experimental evidence benefits LLM-based feasibility assessment and when it introduces fragility.

Large Language Models (LLMs) have demonstrated exceptional performance across diverse domains but are increasingly constrained by high inference latency. Early Exit has emerged as a promising solution to accelerate inference by dynamically bypassing redundant layers. However, in decoder-only architectures, the efficiency of Early Exit is severely bottlenecked by the KV Cache Absence problem, where skipped layers fail to provide the necessary historical states for subsequent tokens. Existing solutions, such as recomputation or masking, either introduce significant latency overhead or incur severe precision loss, failing to bridge the gap between theoretical layer reduction and practical wall-clock speedup. In this paper, we propose River-LLM, a training-free framework that enables seamless token-level Early Exit. River-LLM introduces a lightweight KV-Shared Exit River that allows the backbone's missing KV cache to be naturally generated and preserved during the exit process, eliminating the need for costly recovery operations. Furthermore, we utilize state transition similarity within decoder blocks to predict cumulative KV errors and guide precise exit decisions. Extensive experiments on mathematical reasoning and code generation tasks demonstrate that River-LLM achieves 1.53 to 2.16 times of practical speedup while maintaining high generation quality.

Large language models are rapidly evolving into interactive coding agents capable of end-to-end web coding, yet existing benchmarks evaluate only narrow slices of this capability, typically text-conditioned generation with static-correctness metrics, leaving visual fidelity, interaction quality, and codebase-level reasoning largely unmeasured. We introduce WebCompass, a multimodal benchmark that provides unified lifecycle evaluation of web engineering capability. Recognizing that real-world web coding is an iterative cycle of generation, editing, and repair, WebCompass spans three input modalities (text, image, video) and three task types (generation, editing, repair), yielding seven task categories that mirror professional workflows. Through a multi-stage, human-in-the-loop pipeline, we curate instances covering 15 generation domains, 16 editing operation types, and 11 repair defect types, each annotated at Easy/Medium/Hard levels. For evaluation, we adopt a checklist-guided LLM-as-a-Judge protocol for editing and repair, and propose a novel Agent-as-a-Judge paradigm for generation that autonomously executes generated websites in a real browser, explores interactive behaviors via the Model Context Protocol (MCP), and iteratively synthesizes targeted test cases, closely approximating human acceptance testing. We evaluate representative closed-source and open-source models and observe that: (1) closed-source models remain substantially stronger and more balanced; (2) editing and repair exhibit distinct difficulty profiles, with repair preserving interactivity better but remaining execution-challenging; (3) aesthetics is the most persistent bottleneck, especially for open-source models; and (4) framework choice materially affects outcomes, with Vue consistently challenging while React and Vanilla/HTML perform more strongly depending on task type.

Large Language Models (LLMs) exhibit a puzzling disparity in their formal linguistic competence: while they learn some linguistic phenomena with near-perfect mastery, they often perform below chance on others, even after training on trillions of tokens. In this work, we investigate whether these failures stem from inherent architectural limitations or simply the scarcity of these specific grammatical constructions in web-scale corpora. We pre-train simple GPT-2 Small (124M) models on a 100M-token random sample of the FineWeb corpus and intervene by injecting a minimal amount (1%) of synthetic data targeting specific linguistic phenomena. We find that this targeted intervention substantially improves model performance in 8 out of the 9 worst-performing BLiMP paradigms - notably the accuracy on a specific paradigm, only_npi_scope, surges from 20.9% to 69.4%. Furthermore, we observe that these interventions generally preserve or slightly improve aggregate performance. However, while we also identify a resistant phenomenon, principle_A_c_command, whose performance remains below chance even after our data augmentation, our findings do serve as an optimistic existence proof that even small language models can substantially improve on those linguistic phenomena on which models typically perform poorly, provided the pre-training data contains sufficient exposure to them. This suggests that efforts towards human-scale language modeling may benefit greatly by focusing on data composition. The code to reproduce our results is open-sourced at https://github.com/kowndinya-renduchintala/heterogeneity-in-formal-linguistic-competence.

Large language models (LLMs) offer strong capabilities but raise cost and privacy concerns, whereas small language models (SLMs) facilitate efficient and private local inference yet suffer from limited capacity. To synergize the complementary strengths, we introduce a dynamic collaboration framework, where an SLM learns to proactively decide how to request an LLM during multi-step reasoning, while the LLM provides adaptive feedback instead of acting as a passive tool. We further systematically investigate how collaboration strategies are shaped by SLM and LLM capabilities as well as efficiency and privacy constraints. Evaluation results reveal a distinct scaling effect: stronger SLMs become more self-reliant, while stronger LLMs enable fewer and more informative interactions. In addition, the learned dynamic collaboration strategies significantly outperform static pipelines and standalone inference, and transfer robustly to unseen LLMs.

With the rapid advancement of large language models (LLMs), mobile agents have emerged as promising tools for phone automation, simulating human interactions on screens to accomplish complex tasks. However, these agents often suffer from low accuracy, misinterpretation of user instructions, and failure on challenging tasks, with limited prior work examining why and where they fail. To address this, we introduce DailyDroid, a benchmark of 75 tasks in five scenarios across 25 Android apps, spanning three difficulty levels to mimic everyday smartphone use. We evaluate it using text-only and multimodal (text + screenshot) inputs on GPT-4o and o4-mini across 300 trials, revealing comparable performance with multimodal inputs yielding marginally higher success rates. Through in-depth failure analysis, we compile a handbook of common failures. Our findings reveal critical issues in UI accessibility, input modalities, and LLM/app design, offering implications for future mobile agents, applications, and UI development.

Code secrets are sensitive assets for software developers, and their leakage poses significant cybersecurity risks. While the rapid development of AI code assistants powered by Code Large Language Models (CLLMs), CLLMs are shown to inadvertently leak such secrets due to a notorious memorization phenomenon. This study first reveals that Byte-Pair Encoding (BPE) tokenization leads to unexpected behavior of secret memorization, which we term as \textit{gibberish bias}. Specifically, we identified that some secrets are among the easiest for CLLMs to memorize. These secrets yield high character-level entropy, but low token-level entropy. Then, this paper supports the biased claim with numerical data. We identified that the roots of the bias are the token distribution shift between the CLLM training data and the secret data. We further discuss how gibberish bias manifests under the ``larger vocabulary'' trend. To conclude the paper, we discuss potential mitigation strategies and the broader implications on current tokenizer design.

The explosive growth of large language models (LLMs) has created a heterogeneous and poorly documented ecosystem, making systematic model comparison increasingly important for provenance auditing, security analysis, and model selection. Existing representation methods struggle to address this setting efficiently. Approaches analyzing internal parameters are powerful when architectures are compatible, but face scalability barriers under structural heterogeneity, while methods relying on external outputs may conflate models with similar behaviors and are difficult to align in richer output spaces across different tokenizers. To bridge this gap, we propose ABLE (Attribution-Based Large-model Embedding), a framework that leverages the interpretability space to construct model representations. By aggregating gradient-based feature attributions via a tokenizer-agnostic word-level alignment, ABLE captures model-specific input-sensitivity patterns rather than only surface-level outputs. Beyond empirical utility, we provide a stability analysis showing that, under standard regularity assumptions for differentiable Transformer-style models, ABLE induces a Lipschitz-continuous parameter-to-embedding map with finite-sample convergence guarantees. Extensive experiments on 239 open-source LLMs demonstrate that our training-free approach achieves competitive or superior performance in relation prediction, model routing, and benchmark score prediction.

Interpretability tools are increasingly used to analyze failures of Large Language Models (LLMs), yet prior work largely focuses on short prompts or toy settings, leaving their behavior on commonly used benchmarks underexplored. To address this gap, we study contrastive, LRP-based attribution as a practical tool for analyzing LLM failures in realistic settings. We formulate failure analysis as \textit{contrastive attribution}, attributing the logit difference between an incorrect output token and a correct alternative to input tokens and internal model states, and introduce an efficient extension that enables construction of cross-layer attribution graphs for long-context inputs. Using this framework, we conduct a systematic empirical study across benchmarks, comparing attribution patterns across datasets, model sizes, and training checkpoints. Our results show that this token-level contrastive attribution can yield informative signals in some failure cases, but is not universally applicable, highlighting both its utility and its limitations for realistic LLM failure analysis. Our code is available at: https://aka.ms/Debug-XAI.

Segment-level disclosures are a central component of financial reporting, providing insight into firms' internal organization and the allocation of economic activities across operating units. However, segment information is often presented in both qualitative and quantitative forms, dispersed across tables and narrative sections of Form 10-K filings. Empirical research relying on structured databases faces both completeness and comparability challenges, as some firm-year observations may be missing, nested segment disclosures are not captured, and support for longitudinal and cross-firm comparability is limited. This study develops a large language model-based framework to extract segment disclosures directly from Form 10-K filings and to preserve both reportable and nested segment information. We further design a retrieval augmented system that incorporates information across multiple filings to support comparability. We use two representative settings to demonstrate its application: longitudinal analysis within a firm to interpret segment changes over time, and cross firm alignment of geographic segments across firms with different reporting structures. The results indicate that the artifact accurately extracts segment-level information and effectively addresses questions that require cross-period knowledge, demonstrating the potential of LLM-based approaches to enhance the measurement and interpretation of segment disclosures.

LLMs are increasingly deployed in dynamic, real-world settings, where the distribution of user prompts can shift substantially over time as new tasks, prompts, and users are introduced to a deployed model. Such natural prompt distribution shift poses a major challenge to LLM reliability, particularly for specialized models designed for narrow domains or user populations. Despite attention to out-of-distribution robustness, there is very limited exploration of measuring natural prompt distribution shift in prior work, and its impact on deployed LLMs remains poorly understood. We introduce the LLM Evaluation under Natural prompt Shift (LENS) framework: a data-centric approach for quantifying natural prompt distribution shift and evaluating its effect on the performance of deployed LLMs. We perform a large-scale evaluation using 192 real-world post-deployment prompt shift settings over time, user group, and geographic axes, training a total of 81 models on 4.68M training prompts, and evaluating on 57.6k prompts. We find that even moderate shifts in user prompt behavior correspond with large performance drops (73% average loss) in deployed LLMs. This performance degradation is particularly prevalent when users from different latent groups and geographic regions interact with models and is correlated with natural prompt distribution shift over time. We systematically characterize how LLM instruction following ability degrades over time and between user groups. Our findings highlight the critical need for data-driven monitoring to ensure LLM performance remains stable across diverse and evolving user populations.

This study investigates the phenomenon of hallucinations in domain-adapted Large Language Models (LLMs), focusing on the fine-tuning of the Llama-2 model with the Lamini dataset. Hallucinations, or the generation of nonsensical or unfaithful content by LLMs, pose a significant challenge, especially when these models are fine-tuned with domain-specific data. Our methodology involves a series of experiments testing memorization, recall, and reasoning capabilities of the fine-tuned LLM, comparing its performance on novel question-answer pairs and domain-specific information. We found that while the model shows proficiency in tasks similar to its training data, its capability to accurately reason about and recall new domain-specific information remains limited, leading to instances of hallucination. The model demonstrates a tendency to provide correct answers with extra information, suggesting an inclination toward over-generation. These results suggest important limitations of fine-tuning-only approaches for mitigating hallucinations when adapting LLMs to specialized domains and underscore the need for more robust methods in adapting LLMs to specialized domains. The study also provides insights into the varying performance of LLMs on different types of information, revealing a comparative weakness in handling domain-specific queries.

Multi-hop question answering (MHQA) enables accurate answers to complex queries by retrieving and reasoning over evidence dispersed across multiple documents. Existing MHQA approaches mainly rely on iterative retrieval-augmented generation, which suffer from the following two major issues. 1) Existing methods prematurely commit to surface-level entities rather than underlying reasoning structures, making question decomposition highly vulnerable to lexical ambiguity. 2) Existing methods overlook the logical dependencies among reasoning steps, resulting in uncoordinated execution. To address these issues, we propose STRIDE, a framework that separates strategic planning, dynamic control, and grounded execution. At its core, a Meta-Planner first constructs an entity-agnostic reasoning skeleton to capture the abstract logic of the query, thereby deferring entity grounding until after the reasoning structure is established, which mitigates disambiguation errors caused by premature lexical commitment. A Supervisor then orchestrates sub-question execution in a dependency-aware manner, enabling efficient parallelization where possible and sequential coordination when necessary. By dynamically deciding whether to retrieve new evidence or infer from existing facts, it avoids redundant queries and error propagation, while fusing cross-branch information and reformulating failed queries to enhance robustness. Grounded fact extraction and logical inference are delegated to specialized execution modules, ensuring faithfulness through explicit separation of retrieval and reasoning. We further propose STRIDE-FT, a modular fine-tuning framework that uses self-generated execution trajectories from STRIDE, requiring neither human annotations nor stronger teacher models. Experiments show that STRIDE achieves robust and accurate reasoning, while STRIDE-FT effectively enhances open-source LLMs.

We present a methodological framework to discover linguistic and discursive patterns associated to different social groups through contrastive synthetic text generation and statistical analysis. In contrast with previous approaches, we aim to characterize subtle expressions of bias, instead of diagnosing bias through a pre-determined list of words or expressions. We are also working with contextualized data instead of isolated words or sentences. Our methodology applies to textual productions in any genre, encompassing narrative, task-oriented or dialogic. Contextualized data are generated using controlled combinations of situational scenarios and group markers, creating minimal pairs of texts that differ only in the referenced group while maintaining comparable narrative conditions. To facilitate robust analysis, linguistic forms are generalized and associations between linguistic abstractions and groups are quantified using a variant of pointwise mutual information to detect expressions that appear disproportionately across groups. A fragment-ranking strategy then prioritizes text segments with a high concentration of biased linguistic signals, which allows for experts to assess the harmful potential of linguistic expressions in context, bridging quantitative analysis and qualitative interpretation.

We present, to our knowledge, the most comprehensive cross-model evaluation of LLM agents on offensive cybersecurity tasks, benchmarking 10 frontier models from 7 providers on all 200 challenges of the NYU CTF Bench. Building on the D-CIPHER multi-agent framework, we extend it with multi-provider backend support, a custom Kali Linux environment with over 100 pre-installed penetration testing tools, and runtime tool-discovery agents. Through a controlled factorial study, we find that the Kali Linux environment yields a +9.5 percentage-point improvement over Ubuntu, while auto-prompting and category-specific tips often degrade performance in well-equipped environments. Among models, Claude 4.5 Opus achieves the highest solve rate (59%), followed by Gemini 3 Pro (52%), with Gemini 3 Flash offering the best cost-efficiency at $0.05 per solve. Asymmetric planner/executor model assignments provide no meaningful benefit while coherent same-model configurations consistently outperform mixed-tier pairings. Our results indicate that environment tooling and model selection emerge as the strongest drivers of performance, whereas prompt engineering interventions show diminishing or negative returns in well-equipped environments. Reported performance reflects both model reasoning ability and compatibility with agent tooling and API integration.

Safety-aligned large language models (LLMs) often generate refusal responses to harmless queries due to the over-refusal problem. However, existing methods for mitigating over-refusal cannot maintain a low refusal ratio for harmless queries while keeping a high refusal ratio for malicious ones. In this paper, we analyze how system prompts with varying safety levels affect LLM refusal behaviors when facing over-refusal queries. A key observation is that, when LLMs suffer from the over-refusal issue, non-refusal tokens remain present in the next-token candidate list, but the model systematically fails to select them, despite the generation of refusal tokens. Based on this observation, we propose a training-free and model-agnostic approach, Adaptive Contrastive Decoding (AdaCD), to mitigate over-refusal while maintaining LLM safety. First, AdaCD compares the output distributions of the LLM with or without an extreme safety system prompt to refine the refusal token distribution. Second, we introduce an adaptive contrastive decoding strategy that dynamically incorporates or removes the refusal token distribution, adaptively boosting the probability of selecting refusal or non-refusal tokens. Experimental results on five benchmark datasets show that, on average, AdaCD reduces the refusal ratio for over-refusal queries by 10.35%, yet still increases the refusal ratio for malicious queries by 0.13%. Code is available at https://github.com/OutdoorManofML/AdaCD.

Benchmarking of open-source LLMs for hardware design focuses on which LLMs to use, while treating inference-time decoding configuration as a secondary concern. This work shows that it matters more how an LLM is configured than which model is selected. Benchmarking 26 open-source LLMs on VerilogEval and RTLLM with synthesis-in-the-loop evaluation, the study first maps the current capability landscape and then conducts an extensive 108-configuration hyperparameter sweep on three prominent models. The sweep reveals absolute pass-rate gaps of up to 25.5% between the best and worst settings for the same LLM, which is 5x larger than the average spread observed across various model families under their respective default configurations. Ranking all configurations by Spearman's $ρ$ across the two benchmark suites yields near-zero correlation, demonstrating that optimal configurations do not transfer. These results show that benchmarking conducted under default hyperparameters confounds model capabilities with configuration effects. Realizing the full potential of open-source LLMs for RTL generation requires architecture and benchmark aware hyperparameter selection, as enabled by the proposed methodology.

When does an LLM controller outperform rule-based traversal for knowledge graph exploration? We study this question through RLM-on-KG, a retrieval system that treats an LLM as an autonomous navigator over an RDF-encoded mention graph for grounded question answering. Unlike GraphRAG pipelines that rely on offline LLM indexing, RLM-on-KG performs entity-first, multi-hop exploration at query time using deterministic graph construction and a fixed tool set. Our central finding is a conditional advantage: the value of LLM control depends on evidence scatter and tool-calling sophistication. The paper's core claim is LLM control versus heuristic traversal, not a generic win over GraphRAG. On GraphRAG-Bench Novel (519 questions), Gemini 2.0 Flash achieves +2.47 pp F1 over a rule-based heuristic baseline (p < 0.0001), but only +0.16 pp over a GraphRAG-local variant (not significant). With a stronger controller, Claude Haiku 4.5, the gain over heuristic grows to +4.37 pp (p < 0.001) and extends to a +2.42 pp significant improvement over GraphRAG-local (p < 0.001). The gain is largest when gold evidence is scattered across 6-10 chunks (+3.21 pp) and smallest for concentrated evidence (+1.85 pp). Cross-scale validation on MuSiQue confirms that the LLM-over-heuristic advantage transfers, with expected attenuation on smaller per-question graphs. The core architectural insight is the separation of candidate discovery from ranking: the LLM adds value through exploration breadth, while final evidence selection is best handled by pure vector re-ranking. Beyond retrieval, exploration traces provide a proposed stress-test harness for structured data quality, yielding diagnostics for coverage, connectivity, provenance, and queryability.

Background: Large language models (LLMs) are increasingly proposed for diagnostic support, but few evaluations use real-world multimodal inpatient data, particularly in low and middle-income country (LMIC) public hospitals. Methods: We conducted VALID, a retrospective evaluation of 539 multimodal inpatient cases from a tertiary public hospital in South Africa. Inputs included radiology imaging (CT, MRI, CXR) and reports, laboratory results, clinical notes, and vital signs. Expert panels adjudicated 300 cases (balanced and discordant subsets) to establish ground truth diagnoses, differentials, and reasoning. Ten multimodal LLMs generated zero-shot outputs. A calibrated three-model LLM Jury scored all outputs and routine ward diagnoses across diagnostic accuracy, differential quality, reasoning, and patient safety (>10,000 evaluations). Primary outcomes were composite scores ($S_3$, $S_4$) and win rates. Results: (i) LLM performance was tightly clustered (<15% variation) despite large cost differences; low-cost models performed comparably to top models. (ii) All LLMs significantly outperformed routine ward diagnoses on average diagnostic and safety scores. (iii) Top performance was achieved by GPT-5.1, followed by Gemini models. (vi) Adding radiology reports improved performance by 6%. (v) Diagnostic and reasoning scores were highly correlated ($ρ= 0.85$). (vi) Output rates varied (65-100%) due to input constraints. Results were robust across subsets and evaluation design. Conclusions: Across a real-world LMIC dataset, multimodal LLMs showed similar diagnostic performance despite large cost differences and outperformed routine care on average safety metrics. Affordability, robustness, and deployment constraints may outweigh marginal performance differences in LMIC settings.

Supervised Fine-Tuning (SFT) accelerates taskspecific large language models (LLMs) development, but the resulting proliferation of finetuned models incurs substantial memory overhead. Delta compression addresses this by retaining a single pre-trained LLM with multiple compressed delta weights. However, existing methods fail on models fine-tuned with largescale datasets. We find that larger SFT data scale amplifies delta parameter magnitude, singular values, and entropy, exacerbating compression errors. To tackle this, we propose DQRELO (Delta Compression via Quantization and Residual Low-Rank), a novel training- and data-free delta compression method. It combines coarse-grained one-bit quantization to capture the dominant structure of the delta, followed by compensated residual low-rank approximation to recover fine-grained details from the smaller residual error. Experiments on various LLMs spanning dense and MoE architectures across multiple domains under this challenging setting demonstrate that DQRELO outperforms existing methods. Moreover, we establish key design principles for delta compression through extensive empirical analysis, demonstrating how task difficulty, architecture, and layer positioning create predictable patterns that can guide optimal compression strategies in production systems.

In long-context decoding for LLMs and LMMs, attention becomes increasingly memory-bound because each decoding step must load a large amount of KV-cache data from GPU memory. Existing acceleration strategies often trade efficiency for accuracy by relying on heuristic pruning that may discard useful information. At a deeper level, they also tend to indiscriminately preserve all high-scoring tokens, treat early tokens as indispensable anchors, or rely on heuristic head routing, reflecting an insufficient mechanistic understanding of the attention sink phenomenon. In this paper, we show that the attention sink phenomenon corresponds to a stable, reachable, and error-controllable fixed point constructed during training. Based on this insight, we propose SinkRouter, a training-free selective routing framework that detects the sink signal and skips computations that would otherwise produce near-zero output. To translate this mechanism into real-world acceleration, we develop a hardware-aware Triton kernel with block-level branching and Split-K parallelism. We conduct extensive evaluations on a diverse suite of long-context benchmarks, including LongBench, InfiniteBench, CVBench, MileBench, and MMVP, using both text-only and multimodal backbones such as Llama-3.1-8B, Llama-3.1-70B, Yi-9B-200K, LLaVA-1.5-7B, and LLaVA-1.5-13B. Across these settings, SinkRouter consistently improves decoding efficiency while maintaining competitive accuracy, and reaches 2.03x speedup with a 512K context.

Large language models (LLMs) tuned for safety often avoid acknowledging demographic differences, even when such acknowledgment is factually correct (e.g., ancestry-based disease incidence) or contextually justified (e.g., religious hiring preferences). This identity-blindness yields incorrect responses, unnecessary refusals, or generic "equal-treatment" defaults. We study this via difference-awareness classification: given a question involving demographic groups, the task is not to answer directly, but to classify whether a correct answer requires recognizing group differences (yes) or whether groups should be treated identically (no). Crucially, fine-tuning for accuracy triggers harm drift: model-generated explanations become increasingly harmful as decision accuracy improves, whether by elaborating harmful content, introducing problematic assumptions, or failing to flag harms the baseline identified. To mitigate this, we introduce DART (Distill--Audit--Repair Training), which distills label-conditioned reasoning from a teacher, audits outputs for harm drift cases relative to baseline, and repairs problematic cases via severity-weighted fine-tuning. On eight benchmarks, DART improves Llama-3-8B-Instruct accuracy from 39.0% to 68.8%, with largest gains on equal-treatment prompts (11.3% -> 72.6%), while reducing harm drift cases by 72.6%. It also transfers to 280 open-ended real-world queries across medical, legal, policy, and educational domains, improving difference-appropriate responses from 39.8% to 77.5% while reducing refusals from 34.3% to 3.0%. Our results demonstrate that accuracy and safety need not conflict when explicit detection and repair mechanisms are in place.

Multimodal large language models (MLLMs) have recently demonstrated strong capabilities in understanding and generating responses from diverse visual inputs, including high-resolution images and long video sequences. As these models scale to richer visual representations, inference increasingly relies on storing large numbers of vision tokens in the key-value (KV) cache, making memory consumption a central bottleneck. Existing methods address this issue by identifying redundancy in vision tokens and compressing the cache, but such compression is typically applied only after all inputs are processed, resulting in high peak memory usage during the prefill stage. In this work, we show that MLLMs exhibit inherent structural regularities and representational redundancy that can be exploited to control memory growth throughout inference. Based on this insight, we propose a sequential input-compression mechanism that enforces a fixed memory budget by performing structure-aware key-value cache compression during the prefill process. This approach substantially reduces peak memory usage while maintaining generative performance with only minimal degradation, enabling more practical and memory-efficient multimodal inference.

State-of-the-art large language models (LLMs) show high performance in general visual question answering. However, a fundamental limitation remains: current architectures lack the native 3D spatial reasoning required to directly analyze volumetric medical imaging, such as CT or MRI. Emerging agentic AI offers a new solution, eliminating the need for intrinsic 3D processing by enabling LLMs to orchestrate and leverage specialized external tools. Yet, the feasibility of such agentic frameworks in complex, multi-step radiological workflows remains underexplored. In this work, we present a training-free agentic pipeline for automated brain MRI analysis. Validating our methodology on several LLMs (GPT-5.4, Gemini 3.1 Pro, Claude Sonnet 4.6) with off-the-shelf domain-specific tools, our system autonomously executes complex end-to-end workflows, including preprocessing (skull stripping, registration), pathology segmentation (glioma, meningioma, metastases), and volumetric reasoning. We evaluate our framework across increasingly complex radiological tasks, from single-scan tumor and anatomy analysis to longitudinal response assessment requiring multi-timepoint comparisons. We analyze the impact of architectural design by comparing single-agent models against multi-agent "domain-expert" collaborations. To support rigorous evaluation of future agentic systems, we release a benchmark dataset of image-prompt-answer tuples derived from public data. Our results demonstrate that agentic AI can solve highly neuro-radiological image analysis tasks through tool use without the need for training or fine-tuning.

Traffic prediction plays a central role in intelligent transportation systems (ITS) by supporting real-time decision-making, congestion management, and long-term planning. However, many existing approaches face practical limitations. Most spatio-temporal models are trained on centralized data, rely on numerical representations, and offer limited explainability. Recent Large Language Model (LLM) methods improve reasoning capabilities but typically assume centralized data availability and do not fully capture the distributed and heterogeneous nature of real-world traffic systems. To address these challenges, this study proposes FedLLM (Federated LLM), a privacy-preserving and distributed framework for explainable multi-horizon short-term traffic flow prediction (15-60 minutes). The framework introduces four key contributions: 1) a Composite Selection Score (CSS) for data-driven freeway selection that captures structural diversity across traffic regions 2) a domain-adapted LLM fine-tuned on structured traffic prompts encoding spatial, temporal, and statistical context 3) FedLLM framework, that enables collaborative training across heterogeneous clients while exchanging only lightweight LoRA adapter parameters, 4) a structured prompt representation that supports contextual reasoning and cross-region generalization. The FedLLM design allows each client to learn from local traffic patterns while contributing to a shared global model through efficient parameter exchange, reducing communication overhead and keeping data private. This setup supports learning under non-IID traffic distributions. Experimental results show that FedLLM achieves improved predictive performance over centralized baselines, while producing structured and explainable outputs. These findings highlight the potential of combining FL with domain-adapted LLMs for scalable, privacy-aware, and explainable traffic prediction.

Firearm violence is a pressing public health issue, yet research into survivors' lived experiences remains underfunded and difficult to scale. Qualitative research, including in-depth interviews, is a valuable tool for understanding the personal and societal consequences of community firearm violence and designing effective interventions. However, manually analyzing these narratives through thematic analysis and inductive coding is time-consuming and labor-intensive. Recent advancements in large language models (LLMs) have opened the door to automating this process, though concerns remain about whether these models can accurately and ethically capture the experiences of vulnerable populations. In this study, we assess the use of open-source LLMs to inductively code interviews with 21 Black men who have survived community firearm violence. Our results demonstrate that while some configurations of LLMs can identify important codes, overall relevance remains low and is highly sensitive to data processing. Furthermore, LLM guardrails lead to substantial narrative erasure. These findings highlight both the potential and limitations of LLM-assisted qualitative coding and underscore the ethical challenges of applying AI in research involving marginalized communities.

Hyperspectral tree species classification is challenging due to limited and imbalanced class labels, spectral mixing (overlapping light signatures from multiple species), and ecological heterogeneity (variability among ecological systems). Addressing these challenges requires methods that integrate biological and structural characteristics of vegetation, such as canopy architecture and interspecific interactions, rather than relying solely on spectral signatures. This paper presents a biologically informed, semi-supervised deep learning method that integrates multi-sensor Earth observation data, specifically hyperspectral imaging (HSI) and airborne laser scanning (ALS), with expert, ecological knowledge. The approach relies on biologically inspired pseudo-labelling over a precomputed canopy graph, yielding accurate classification at low training cost. In addition, ecological priors on species cohabitation are automatically derived from reliable sources using large language models (LLMs) and encoded as a cohabitation matrix with likelihoods of species occurring together. These priors are incorporated into the pseudo-labelling strategy, effectively introducing expert knowledge into the model. Experiments on a real-world forest dataset demonstrate 5.6% improvement over the best reference method. Expert evaluation of cohabitation priors reveals high accuracy with differences no larger than 15%.

Real-effort tasks, in which participants perform cognitively costly activities whose outcomes depend on actual performance, are widely used in experimental economics. Their validity, however, rests on the assumption that a human performs them. We study whether this assumption still holds in the era of Artificial Intelligence (AI) and Large Language Models (LLMs). Using 8 canonical real-effort tasks and 23 LLMs from three major providers, we show that most tasks can now be solved accurately and at a negligible cost, while only a few resist automation. Performance improves with each model generation, and midtier models are rapidly closing the gap with frontier ones, broadening the set of widely accessible models that can automate these tasks. Additionally, we show that verbally offering monetary incentives has no effect on LLM performance. Our findings establish a boundary condition for the use of real-effort tasks in unsupervised settings: when participants can cheaply outsource task completion to an LLM, observed performance may no longer reflect genuine human effort.

Large language models (LLMs) have demonstrated impressive capabilities, yet their internal mechanisms for handling reasoning-intensive tasks remain underexplored. To advance the understanding of model-internal processing mechanisms, we present an investigation of how LLMs perform arithmetic operations by examining internal mechanisms during task execution. Using early decoding, we trace how next-token predictions are constructed across layers. Our experiments reveal that while the models recognize arithmetic tasks early, correct result generation occurs only in the final layers. Notably, models proficient in arithmetic exhibit a clear division of labor between attention and MLP modules, where attention propagates input information and MLP modules aggregate it. This division is absent in less proficient models. Furthermore, successful models appear to process more challenging arithmetic tasks functionally, suggesting reasoning capabilities beyond factual recall.

Clinicians exploring oncology trial repositories often need ad-hoc, multi-constraint queries over biomarkers, endpoints, interventions, and time, yet writing SQL requires schema expertise. We demo FD-NL2SQL, a feedback-driven clinical NL2SQL assistant for SQLite-based oncology databases. Given a natural-language question, a schema-aware LLM decomposes it into predicate-level sub-questions, retrieves semantically similar expert-verified NL2SQL exemplars via sentence embeddings, and synthesizes executable SQL conditioned on the decomposition, retrieved exemplars, and schema, with post-processing validity checks. To improve with use, FD-NL2SQL incorporates two update signals: (i) clinician edits of generated SQL are approved and added to the exemplar bank; and (ii) lightweight logic-based SQL augmentation applies a single atomic mutation (e.g., operator or column change), retaining variants only if they return non-empty results. A second LLM generates the corresponding natural-language question and predicate decomposition for accepted variants, automatically expanding the exemplar bank without additional annotation. The demo interface exposes decomposition, retrieval, synthesis, and execution results to support interactive refinement and continuous improvement.

Adaptive Retrieval-Augmented Generation aims to mitigate the interference of extraneous noise by dynamically determining the necessity of retrieving supplementary passages. However, as Large Language Models evolve with increasing robustness to noise, the necessity of adaptive retrieval warrants re-evaluation. In this paper, we rethink this necessity and propose AdaRankLLM, a novel adaptive retrieval framework. To effectively verify the necessity of adaptive listwise reranking, we first develop an adaptive ranker employing a zero-shot prompt with a passage dropout mechanism, and compare its generation outcomes against static fixed-depth retrieval strategies. Furthermore, to endow smaller open-source LLMs with this precise listwise ranking and adaptive filtering capability, we introduce a two-stage progressive distillation paradigm enhanced by data sampling and augmentation techniques. Extensive experiments across three datasets and eight LLMs demonstrate that AdaRankLLM consistently achieves optimal performance in most scenarios with significantly reduced context overhead. Crucially, our analysis reveals a role shift in adaptive retrieval: it functions as a critical noise filter for weaker models to overcome their limitations, while serving as a cost-effective efficiency optimizer for stronger reasoning models.

Large Language Models (LLMs) are poised to disrupt knowledge work, with the emergence of delegated work as a new interaction paradigm (e.g., vibe coding). Delegation requires trust - the expectation that the LLM will faithfully execute the task without introducing errors into documents. We introduce DELEGATE-52 to study the readiness of AI systems in delegated workflows. DELEGATE-52 simulates long delegated workflows that require in-depth document editing across 52 professional domains, such as coding, crystallography, and music notation. Our large-scale experiment with 19 LLMs reveals that current models degrade documents during delegation: even frontier models (Gemini 3.1 Pro, Claude 4.6 Opus, GPT 5.4) corrupt an average of 25% of document content by the end of long workflows, with other models failing more severely. Additional experiments reveal that agentic tool use does not improve performance on DELEGATE-52, and that degradation severity is exacerbated by document size, length of interaction, or presence of distractor files. Our analysis shows that current LLMs are unreliable delegates: they introduce sparse but severe errors that silently corrupt documents, compounding over long interaction.

Existing research on large language models (LLMs) for automated code compliance has primarily focused on performance, treating the models as black boxes and overlooking how training decisions affect their interpretive behavior. This paper addresses this gap by employing a perturbation-based attribution analysis to compare the interpretive behaviors of LLMs across different fine-tuning strategies such as full fine-tuning (FFT), low-rank adaptation (LoRA) and quantized LoRA fine-tuning, as well as the impact of model scales which include varying LLM parameter sizes. Our results show that FFT produces attribution patterns that are statistically different and more focused than those from parameter-efficient fine-tuning methods. Furthermore, we found that as model scale increases, LLMs develop specific interpretive strategies such as prioritizing numerical constraints and rule identifiers in the building text, albeit with performance gains in semantic similarity of the generated and reference computer-processable rules plateauing for models larger than 7B. This paper provides crucial insights into the explainability of these models, taking a step toward building more transparent LLMs for critical, regulation-based tasks in the Architecture, Engineering, and Construction industry.

LLMbench is a browser-based workbench for the comparative close reading of large language model (LLM) outputs. Where existing tools for LLM comparison, such as Google PAIR's LLM Comparator are engineered for quantitative evaluation and user-rating metrics, LLMbench is oriented towards the hermeneutic practices of the digital humanities. Two model responses to the same prompt are side by side in annotatable panels with four analytical overlays (Probabilities for token-level log-probability inspection, Differences for word-level diff across the two panels, Tone for Hyland-style metadiscourse analysis, and Structure for sentence-level parsing with discourse connective highlighting), alongside five analytical modes, Stochastic Variation, Temperature Gradient, Prompt Sensitivity, Token Probabilities, and Cross-Model Divergence, that make the probabilistic structure of generated text legible at the token level. The tool treats the generated text as a research object in its own right from a probability distribution, a text that could have been otherwise, and provides visualisations including continuous heatmaps, entropy sparklines, pixel maps, and three-dimensional probability terrains, that show the counterfactual history from which each word emerged. This paper describes the tool's architecture, its six modes, and its design rationale, and argues that log-probability data, currently underused in humanistic and social-scientific readings of AI, is an important resource for a critical studies of generative AI models.

LLM-based simulators offer a promising path for generating complex synthetic data where traditional differentially private (DP) methods struggle with high-dimensional user profiles. But can LLMs faithfully reproduce statistical distributions from DP-protected inputs? We evaluate this using PersonaLedger, an agentic financial simulator, seeded with DP synthetic personas derived from real user statistics. We find that PersonaLedger achieves promising fraud detection utility (AUC 0.70 at epsilon=1) but exhibits significant distribution drift due to systematic LLM biases--learned priors overriding input statistics for temporal and demographic features. These failure modes must be addressed before LLM-based methods can handle the richer user representations where they might otherwise excel.

Continuous glucose monitoring (CGM) is central to diabetes care, but explaining CGM patterns clearly and empathetically remains time-intensive. Evidence for retrieval-grounded large language model (LLM) systems in CGM-informed counseling remains limited. To evaluate whether a retrieval-grounded LLM-based conversational agent (CA) could support patient understanding of CGM data and preparation for routine diabetes consultations. We developed a retrieval-grounded LLM-based CA for CGM interpretation and diabetes counseling support. The system generated plain-language responses while avoiding individualized therapeutic advice. Twelve CGM-informed cases were constructed from publicly available datasets. Between Oct 2025 and Feb 2026, 6 senior UK diabetes clinicians each reviewed 2 assigned cases and answered 24 questions. In a blinded multi-rater evaluation, each CA-generated and clinician-authored response was independently rated by 3 clinicians on 6 quality dimensions. Safety flags and perceived source labels were also recorded. Primary analyses used linear mixed-effects models. A total of 288 unique responses (144 CA and 144 clinician) generated 864 ratings. The CA received higher quality scores than clinician responses (mean 4.37 vs 3.58), with an estimated mean difference of 0.782 points (95% CI 0.692-0.872; P<.001). The largest differences were for empathy (1.062, 95% CI 0.948-1.177) and actionability (0.992, 95% CI 0.877-1.106). Safety flag distributions were similar, with major concerns rare in both groups (3/432, 0.7% each). Retrieval-grounded LLM systems may have value as adjunct tools for CGM review, patient education, and preconsultation preparation. However, these findings do not support autonomous therapeutic decision-making or unsupervised real-world use.

Frontier model developers aim to train models continually to possess emergent, diverse capabilities. To extend capabilities, the current pre-training and post-training paradigm requires manually starting training runs with static datasets or reward functions every time. Addressing this limitation, our work pursues the insight that open-endedness (via the coevolution of models and tasks) can discover models with increasingly novel skills in a single run. We introduce a new model development framework that extends coevolution to large language model (LLM) discovery, open-ended \textit{Assessment Coevolving with Diverse Capabilities} (AC/DC). AC/DC evolves both LLMs via model merging and natural language tasks via synthetic data generation. AC/DC discovers growing archives of LLMs that surpass the capabilities of larger LLMs while taking up less GPU memory. In particular, our LLM populations achieve a broader Coverage of expertise than other curated models or baselines on downstream benchmarks, without \textit{any} explicit benchmark optimization. Furthermore, AC/DC improves Coverage over time, continually innovates on tasks and models, and improves performance in multi-agent best-of-N selection. Our findings highlight the potential of coevolution as a means of discovering broader sets of capabilities from base LLMs. Overall, AC/DC brings us one step closer to a profoundly new paradigm of LLM development, where continual improvements to the diversity of model capabilities can be accelerated by leveraging existing models as stepping stones to increasingly powerful models.

Test-time compute scaling, the practice of spending extra computation during inference via repeated sampling, search, or extended reasoning, has become a powerful lever for improving large language model performance. Yet deploying these techniques under finite inference budgets requires a decision that current systems largely ignore: which inputs deserve more compute, and which can be answered cheaply? We formalize this as a constrained optimization problem (maximize expected accuracy subject to an average compute budget) and solve it with a two-stage Solve-then-Learn pipeline. In the solve stage, Lagrangian relaxation decomposes the global constraint into per-instance sub-problems, each admitting a closed-form oracle action that optimally prices accuracy against cost. We prove that the induced cost is monotone in the dual variable, enabling exact budget targeting via binary search. In the learn stage, a lightweight classifier is trained to predict oracle actions from cheap input features, amortizing the allocation rule for real-time deployment. We establish that the task-level regret of the learned policy is bounded by its imitation error times the worst-case per-instance gap, yielding a clean reduction from constrained inference to supervised classification. Experiments on MATH and GSM8K with three LLMs (DeepSeek-V3, GPT-4o-mini, Qwen2.5-7B) show that our method consistently outperforms uniform and heuristic allocation baselines, achieving up to 12.8% relative accuracy improvement on MATH under matched budget constraints, while closely tracking the Lagrangian oracle upper bound with over 91% imitation accuracy.

The rapid integration of large language models (LLMs) into everyday workflows has transformed how individuals perform cognitive tasks such as writing, programming, analysis, and multilingual communication. While prior research has focused on model reliability, hallucination, and user trust calibration, less attention has been given to how LLM usage reshapes users' perceptions of their own capabilities. This paper introduces the LLM fallacy, a cognitive attribution error in which individuals misinterpret LLM-assisted outputs as evidence of their own independent competence, producing a systematic divergence between perceived and actual capability. We argue that the opacity, fluency, and low-friction interaction patterns of LLMs obscure the boundary between human and machine contribution, leading users to infer competence from outputs rather than from the processes that generate them. We situate the LLM fallacy within existing literature on automation bias, cognitive offloading, and human-AI collaboration, while distinguishing it as a form of attributional distortion specific to AI-mediated workflows. We propose a conceptual framework of its underlying mechanisms and a typology of manifestations across computational, linguistic, analytical, and creative domains. Finally, we examine implications for education, hiring, and AI literacy, and outline directions for empirical validation. We also provide a transparent account of human-AI collaborative methodology. This work establishes a foundation for understanding how generative AI systems not only augment cognitive performance but also reshape self-perception and perceived expertise.

Large Language Models (LLMs) achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains poorly understood. By analyzing model-generated reasoning trajectories, we find that errors are not uniformly distributed but often originate from a small number of early transition points, after which reasoning remains locally coherent but globally incorrect. These transitions coincide with localized spikes in token-level entropy, and alternative continuations from the same intermediate state can still lead to correct solutions. Based on these observations, we introduce GUARD, a targeted inference-time framework that probes and redirects critical transitions using uncertainty signals. Empirical evaluations across multiple benchmarks confirm that interventions guided by these failure dynamics lead to more reliable reasoning outcomes. Our findings highlight the importance of understanding when and how reasoning first deviates, complementing existing approaches that focus on scaling inference-time computation.

Robust explanations are increasingly required for user trust in enterprise NLP, yet pre-deployment validation is difficult in the common case of black-box deployment (API-only access) where representation-based explainers are infeasible and existing studies provide limited guidance on whether explanations remain stable under real user noise, especially when organizations migrate from encoder classifiers to decoder LLMs. To close this gap, we propose a unified black-box robustness evaluation framework for token-level explanations based on leave-one-out occlusion, and operationalize explanation robustness with top-token flip rate under realistic perturbations (swap, deletion, shuffling, and back-translation) at multiple severity levels. Using this protocol, we conduct a systematic cross-architecture comparison across three benchmark datasets and six models spanning encoder and decoder families (BERT, RoBERTa, Qwen 7B/14B, Llama 8B/70B; 64,800 cases). We find that decoder LLMs produce substantially more stable explanations than encoder baselines (73% lower flip rates on average), and that stability improves with model scale (44% gain from 7B to 70B). Finally, we relate robustness improvements to inference cost, yielding a practical cost-robustness tradeoff curve that supports model and explanation selection prior to deployment in compliance-sensitive applications.

Current LLM systems are increasingly equipped with a code interpreter that executes generated code to obtain results. This works serially: the model first generates the complete code, then an interpreter executes it. This sequential workflow leaves the executor idle during generation and the generator idle during execution, resulting in unnecessary end-to-end latency. Our key observation is that an LLM, unlike a human developer, emits code tokens left to right and does not backtrack over what it has already written. This makes it possible to start executing a piece of code while later tokens are still being generated. We formalize this parallel execution paradigm, modeling it as a three-stage pipeline of generation, detection, and execution, and derive closed-form latency bounds that characterize its speedup potential and operating regimes. We then present EAGER, a concrete implementation featuring AST-based chunking, dynamic batching with gated execution, and early error interruption. We evaluate EAGER across four benchmarks, seven LLMs, and three execution environments. The overlap mechanism hides almost all execution behind generation, reducing the non-overlapped portion of execution time by up to 99.8% and cutting end-to-end latency by up to 37.3% on error-free runs.

Aligned language models that are trained to refuse harmful requests also exhibit over-refusal: they decline safe instructions that seemingly resemble harmful instructions. A natural approach is to ablate the global refusal direction, steering the hidden-state vectors away or towards the harmful-refusal examples, but this corrects over-refusal only incidentally while disrupting the broader refusal mechanism. In this work, we analyse the representational geometry of both refusal types to understand why this happens. We show that harmful-refusal directions are task-agnostic and can be captured by a single global vector, whereas over-refusal directions are task-dependent: they reside within the benign task-representation clusters, vary across tasks, and span a higher-dimensional subspace. Linear probing suggests that the two refusal types are representationally distinct from the early transformer layers. These findings provide a mechanistic explanation of why global direction ablation alone cannot address over-refusal, and establish that task-specific geometric interventions are necessary.

In the maintenance of complex systems, fault trees are used to locate problems and provide targeted solutions. To enable fault trees stored as images to be directly processed by large language models, which can assist in tracking and analyzing malfunctions, we propose a novel textual representation of fault trees. Building on it, we construct a benchmark for multi-turn dialogue systems that emphasizes robust interaction in complex environments, evaluating a model's ability to assist in malfunction localization, which contains $3130$ entries and $40.75$ turns per entry on average. We train an end-to-end model to generate vague information to reflect user behavior and introduce long-range rollback and recovery procedures to simulate user error scenarios, enabling assessment of a model's integrated capabilities in task tracking and error recovery, and Gemini 2.5 pro archives the best performance.

Epistemic reasoning requires agents to infer the state of the world from partial observations and information about other agents' knowledge. Prior work evaluating LLMs on epistemic puzzles often frames failures as memorization rather than reasoning. We argue that this dichotomy is too coarse for newer models: memorization is a limiting case of pattern-based reasoning, where a model matches a task to a familiar template and applies the corresponding solution. We introduce a two-dimensional benchmark over DEL-style puzzles, separating narrative familiarity from inference complexity, allowing us to distinguish pattern-based from epistemic reasoning. We find that models are substantially more robust to surface form changes than prior work suggested, yet consistently struggle in asymmetric settings where familiar patterns no longer apply and success requires tracking fragmented epistemic states.

A number of scientific conferences and journals have recently enacted policies that prohibit LLM usage by peer reviewers, except for polishing, paraphrasing, and grammar correction of otherwise human-written reviews. But, are these policies enforceable? To answer this question, we assemble a dataset of peer reviews simulating multiple levels of human-AI collaboration, and evaluate five state-of-the-art detectors, including two commercial systems. Our analysis shows that all detectors misclassify a non-trivial fraction of LLM-polished reviews as AI-generated, thereby risking false accusations of academic misconduct. We further investigate whether peer-review-specific signals, including access to the paper manuscript and the constrained domain of scientific writing, can be leveraged to improve detection. While incorporating such signals yields measurable gains in some settings, we identify limitations in each approach and find that none meets the accuracy standards required for identifying AI use in peer reviews. Importantly, our results suggest that recent public estimates of AI use in peer reviews through the use of AI-text detectors should be interpreted with caution, as current detectors misclassify mixed reviews (collaborative human-AI outputs) as fully AI generated, potentially overstating the extent of policy violations.