Answer

Follow this topic Ask about this topic

Momentum

24 papers in the last four weeks, up 50% on the four weeks before. 0.3% of all new papers.

Jul 6Week of Sep 21

Latest papers 181

Aug 2, 2026cs.CL

Why LLMs Give In: Conversational Factors and Reasoning Behind Medical Sycophancy

Large language models can answer a medical question correctly and still abandon that answer when a user pushes back. We study this failure as medical sycophancy and ask when models are most likely to give in. Across five open-weight models, 500 MedQuAD questions, and 1.2 million trials, we use a fully crossed design over four conversational factors: user role, user evidence, interaction structure, and grounding. Medical sycophancy is nearly three times more common when users challenge an answer the model has already given than when the false claim appears in the initial query. Models are also more susceptible to users presented as physicians or medical students. Most strikingly, fabricated evidence has opposite effects across interaction structures. It increases sycophancy in single-turn interactions but reduces it after the model has already answered. Grounding helps, but does not eliminate the behavior. Sycophancy varies more across medical questions than across models, making question selection an important part of benchmark design. Reasoning traces suggest that multi-turn failures are associated with models turning back toward their own prior answer, while fabricated evidence receives more scrutiny after an initial response. Together, the results show that medical sycophancy depends as much on how a model is challenged and evaluated as on which model is tested.

Kaike Ping, Buse Çarık, Caleb Wohn +3

Virginia Tech · Shanghai Tongren Hospital, Shanghai Jiao Tong University School of Medicine · Emory University

Sycophancy Confabulation Answer Conversational Context

Jul 31, 2026cs.CL

Sixteen models, fewer than two voices: measuring ensemble dispersion where no answer is uniquely correct

Sixteen language models drawn from ten families produced, on average, the semantic diversity of 1.69 distinct formulations of a psychotherapeutic case, against a single-model baseline of 1.43 from one model's own runs. Ensembles place more than one reading before a decision-maker on the premise that several models supply several perspectives. Dispersion over their outputs is measured both as diversity and as uncertainty, and both traditions validate it against a correctness criterion that this task does not admit. Measuring diversity is a solved problem: the Vendi Score, the exponential of the von Neumann entropy of a similarity matrix, is an effective number of distinct elements. What a single aggregate does not say is where the diversity comes from. We define a per-model dissent contribution, the complement of a model's mean similarity to the other members of its ensemble: a magnitude from the same matrix, not a decomposition of the spectral index, whose maximum identifies the most divergent voice. Crossing model and case, we test as a preregistered hypothesis whether model identity accounts for a non-zero share of the variance in dissent, and characterise the structure that test detects. The panel formulated fifteen stratified vignettes, yielding 7,082 formulations for analysis. Model identity was a detectable structuring factor of the dissent that remained, but the usual categories recovered it only partly: scale differences pointed in opposite directions across pairs, family grouped models on only five two-member lines, and the most divergent voice changed with panel composition, so that the surfaced outlier describes the ensemble rather than the model. Dissent did not track the interpretive openness for which the case bank was stratified; it was organised by clinical content instead, leaving the dispersion an ensemble produces a property to measure rather than assume.

Mario Vega-Barbas, Lidia Mora-Valenciano, Iván Pau +2

ETSIS de Telecomunicación, Universidad Politécnica de Madrid, Madrid, Madrid, Spain · Independent Researcher, Guadalajara, Spain · Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Huddinge, Stockholm, Sweden +4

Diversity Dispersion Disagreement Heterogeneous Modeling Tools Interpretive Layer Ensemble+2

Jul 31, 2026cs.CV

HierDoc: Hierarchical Page-to-Region Evidence Routing for Long-Document Visual Question Answering

Multi-page document visual question answering requires locating sparse evidence at both the page and region levels. Existing approaches typically emphasize one level over the other: page-centric methods focus on page acquisition, with region operations serving mainly as navigation aids, whereas region-centric methods assume that the relevant pages have already been supplied. Consequently, page and region selection remain disconnected rather than forming successive evidence decisions. We propose HierDoc, a hierarchical evidence-routing framework that formulates long-document evidence acquisition as two-stage set prediction from pages to regions. A page policy selects evidence pages from the full document; these pages are then parsed for semantic elements, after which a region policy selects the elements passed to a downstream answer model. Both answer-agnostic policies are optimized with stage-wise GRPO using granularity-specific structured-set rewards. The answer model receives selected full pages together with selected region crops and OCR or table text, preserving global context while emphasizing fine-grained evidence. Across the evaluated benchmarks, HierDoc achieves state-of-the-art or competitive performance among open-weight systems, improving LongDocURL by 16.87% relative to the strongest reported open-weight baseline. Controlled ablations further show that selected regional evidence improves the page-only system in accuracy and F1 by 5.51% and 4.82%, respectively. These results demonstrate the benefit of organizing coarse page routing and fine-grained region routing as successive, separately optimized stages of a unified evidence-acquisition process.

Rongjian Gu, Wengang Zhou, Junyu Xiong +4

University of Science and Technology of China · 2iFLYTEK Research · 3Anhui Transport Consulting & Design Institute Co. Ltd., Hefei, China

Visual Document Retrieval Answer

Jul 31, 2026cs.AI

AMTFV: Agentic Mathematical Tool-Flow Verification for LLM Self-Correction

Large language models have demonstrated strong mathematical problem-solving capabilities, yet reliably verifying their candidate answers remains challenging. Existing representative methods mainly revise outputs through natural-language reflection or assist verification by directly generating verification programs; the former may not reliably support exact computation, whereas the latter prematurely couples mathematical modeling with low-level implementation. We propose AMTFV (Agentic Mathematical Tool-Flow Verification). By introducing Mathematical Tool Flow (MTF) as an interrupt--execute--resume interface, AMTFV decouples verification modeling from concrete execution and supports exact computation through a mathematical toolbox. Specifically, the verification agent first constructs a verification workflow, encodes the mathematical objects and computational intent requiring reliable execution in an MTF request, and sends it to the mathematical toolbox agent. The latter parses the request, generates executable calls, and dispatches them to the backend for exact computation. Tool outputs then support candidate-answer adjudication, answer revision, and verification-workflow revision. We evaluate AMTFV on five challenging mathematical reasoning datasets with seven model configurations from DeepSeek, GPT, and Gemini. Experimental results show that AMTFV outperforms the representative baselines evaluated in this study overall; under an individual model configuration, it improves average accuracy over the strongest baseline by up to 8.3 percentage points, with larger gains on samples of medium and high verification complexity.

Rui Zou, Yutao Zhu, Mengqi Wei +1

Gaoling School of Artificial Intelligence, Renmin University of China · National Engineering Research Center of Educational Big Data, Central China Normal University

LLM Reasoning Strategies Answer

Jul 29, 2026cs.LG

RLPF: Reinforcement Learning from Performance Feedback for Code Generation

Code models are increasingly trained with execution feedback, but most training signals still stop at correctness. This leaves an important gap for systems code: two programs can pass the same tests while differing greatly in runtime. We study how to train code agents to prefer faster correct implementations, rather than treating efficiency only as an evaluation metric. The key difficulty is that runtime is a fragile reward. It is meaningful only after a program is correct, varies across tasks, and gives little guidance when most sampled programs fail to compile or run. We propose \textbf{RLPF}, reinforcement learning from performance feedback, which turns execution outcomes into a staged reward. Failed programs are ordered by execution progress, while correct programs are ranked by their relative improvement from the baseline toward the expert reference. This gives useful feedback before correctness and performance-sensitive feedback after correctness. Fine-tuning Qwen3-32B with RLPF on PerfCodeBench raises correct-and-runnable solutions from

11.1\%

to

54.6\%

and improves relative efficiency from

8.1\%

to

38.6\%

. The trained model becomes competitive with stronger open-weight systems, and its optimization behavior transfers modestly to EffiBench-X. Additional studies show that model-generated references provide useful but weaker supervision, and that the full composite reward is more reliable than correctness-only or runtime-only baselines. These results suggest that code agents can be trained not only to pass tests, but also to optimize the programs they write.

Huihao Jing, Haozhe Cui, Wenbin Hu +9

The Hong Kong University of Science and Technology (HKUST)

Code Generation Feedback Prompt Optimization Answer

Jul 27, 2026cs.LG

GAUGE: Grading Agent-Built Financial Models Without a Golden Answer

Financial models combine public disclosures with analyst assumptions to produce forecasts and valuations. While some components can be checked mechanically, forecasts, discount rates, and target prices often admit multiple reasonable answers. Existing benchmarks nevertheless tend to grade such outputs against a single expert reference. Using independently built analyst models for the same companies, we find that across 108 directed pairs covering 65 companies, the median single-reference score is 0.33, 92.6% score below 0.70, and no same-vintage pair agrees on implied price within 10%. Point-tolerance grading can therefore penalize disagreement already present among professionals. We introduce GAUGE, a benchmark for evaluating agent-built valuation models against observed analyst practice rather than a single point answer. GAUGE uses 1,001 vendor-classified analyst workbooks and a 196-task evaluation set, with a three-layer observed-practice envelope, 56 auditable facets, eight validity gates, and deterministic structural checks. We validate the benchmark with a 55-participant known-groups study, company-grouped cross-fitting, and judge-stability audits. On the failure-aware score

φ_0

, senior analysts average 88.3, juniors 66.0, and finance students 43.2. Across 24 agents and 1,011 scored generations, the best agent scores 53.4, above the student mean but below every senior and most juniors. It passes 93% of mechanical facets and 78% of judgment facets, with a fleet-median gap of 26 points. Current agents are substantially stronger at model construction than valuation judgment. We release the methodology, a gated de-identified data tier, a controlled training split, a versioned 48-task evaluation core, and a withheld refresh pool.

Jiacheng Lu, Sinuo Wang, Wentao Zhao +12

Shanghai Jiao Tong University Shanghai, China · StepFun Shanghai, China · University of Adelaide Adelaide, Australia +5

Finance Benchmarks Grading Agentic Evaluations Value Answer

Jul 23, 2026cs.AI

Bound-Founded Semantics for Answer Set Programming with Difference Constraints: Preliminary Report

While the integration of linear constraints has significantly expanded the reach of Answer Set Programming (ASP), existing hybrid solvers often rely on disparate semantic underpinnings that lack a unified logical foundation. We address this gap by introducing a many-sorted variant of the Bound-founded Logic of Here-and-There (HTb), providing a versatile framework capable of characterizing equilibrium models across a wide spectrum of alternative semantics for extensions of ASP with linear constraints. We apply this framework to the setting of difference constraints, focusing on the semantic characterization of clingo[DL]. Central to our approach is the formalization of foundedness for numeric variables. By investigating how different hybrid systems - such as clingo[DL], clingcon, and flingo - justify constraint atoms, we uncover the semantic roots of their varying behaviors. This investigation results in a single, consistent framework that not only formalizes the foundations of current systems like clingo[DL] but also facilitates the rigorous study of program simplifications and the future integration of diverse semantic principles.

Pedro Cabalar, Jorge Fandinno, Nicolas Rühling +3

University of A Corunna, Spain · University of A Coruña, Spain · University of Nebraska at Omaha, USA +3

Answer Set Programming Inequality Constraints First-Order Logic Answer Neural Solvers

Jul 21, 2026cs.CL

Verifiable Self-Evolution for Open-Ended Dialogue Skills via Future-Feedback Prediction

Textual skills provide a lightweight way to improve frozen language-model agents, but their self-evolution normally requires a stable validation signal. Such signals are natural in mathematics or code, where an answer can be checked after it changes, yet are problematic in open-ended dialogue: changing the assistant response also changes the user's next reaction, so a logged reaction cannot directly evaluate a counterfactual response. We propose future-feedback skill evolution, which first redirects self-evolution from prescribing the current answer to predicting whether the observed answer will lead to a positive or negative subsequent user signal. This prediction task is verifiable on fixed logged tuples and therefore supports validation-gated textual optimization. The evolved feedback skill captures interpretable criteria for response quality and can subsequently serve as a diagnostic and optimization target for answer skills. On a proprietary, privacy-preserving sales-assistant dataset, careful quality filtering and a balanced resolved/unresolved split yield more than 75% prediction accuracy. Beyond this result, the central contribution is a formulation that converts otherwise moving conversational feedback into a fixed offline learning target, enabling reproducible skill evolution without placing every candidate skill in live traffic. We discuss the boundary between observational verification and counterfactual validity, and position the method as an offline optimization stage rather than a replacement for final human or online evaluation.

ChaoJin Zhao, Xuan Jiang

ByteDance

Self-Evolving Skill Libraries Self-Evolution Skills Answer

Jul 20, 2026cs.CL

Structured Output Collapses Answer Diversity Across 44 Language Models

When a language model must choose one answer from a large space of equally valid options, a format clause -- "Reply with JSON only" -- changes which answer it chooses. We re-run the One-Word Census (arXiv:2607.12796): 31 wide-answer-space category prompts asked of 44 models, now with the reply requested in JSON -- no schema enforcement, no constrained decoding, only the request. Convergence deepens sharply: on the unconstrained "Pick a word" prompt the modal answer rises from 41% to 64% of the pool and distinct answers fall from 52 to 36; mean answer-choice surprisal drops from 1.80 to 1.58 bits. The tax is progressive: six of 44 models move individually (BH-FDR q=.10), all toward the mode, led by the most distinctive models, while the conformist floor is immobile. It is a sharpener, not a re-indexer -- the plain-chat modal answer survives in 28 of 31 categories. Defaults are register-indexed: a within-run re-sample (n=20) finds JSON shifts 53% of a model's stable chat defaults, mostly back to the crowd, and installs defaults absent from chat (Claude Fable 5 answers "cerulean" for colour 0% of the time in chat, 100% in JSON). Full-battery controls reveal a register gradient: compression is significant and specific to the answer-delivery formats models are trained to speak (JSON -0.22 bits, p=.0002; XML -0.19, p=.002), absent for YAML and CSV, and reversed for an arbitrary bracket wrapper (+0.13, p=.009) -- weighing the mechanism toward tool-use post-training. Enforcing the schema at the decoder (response_format) compresses no further than the request (-0.03 bits): the collapse lives in the model's response to the register, not the decoder. Structured output is how software consumes language models, and that surface is served by a measurably more homogeneous model than the chat surface on which models are evaluated, compared, and chosen.

Tapan Parikh

Cornell Tech

Large Language Model Responses Answer Language Model Computes Chatbots Diversity Surprisal

Jul 20, 2026cs.AI

Semantically Similar, Yet Not Answerable: Diagnosing the Semantic-Answerability Gap in Table RAG

In retrieval-augmented generation (RAG), semantic relevance asks whether a source matches a query in meaning, while answerability asks whether it contains sufficient information to answer the query. A Semantic-Answerability Gap (SAG) may arise in retrieval when a retriever can reach semantically relevant sources yet fail to identify those that are uniquely answerable. We uncover this gap using tables as a controlled setting, where shared schemas and entities provide strong semantic signals while localized content and row-column bindings distinguish answerable from non-answerable sources. Using TCR-Bench, a controlled sibling-table benchmark, we find that dense retrievers achieve only 18.2% top-1 target retrieval, reducing QA F1 from 0.755 with the oracle table to 0.330 with retrieved top-5 tables. Controlled diagnostics show that retrievers favor semantic volume over sufficiency, respond weakly to row-column binding disruptions, and struggle to distinguish Targets from Siblings. Explicit answerability assessment substantially improves target identification, while fine-tuning shows that answerability is learnable but difficult to transfer without compromising broad semantic retrieval.

Jiaming Tian, Liyao Li, Wentao Ye +6

Zhejiang University · Bank of Hangzhou Co., Ltd. · Zhejiang Lab

Hievi-Rag Tabular Data Answer Semantic Gap Contextual Cues

Jul 19, 2026cs.LG

DRNOISE: Benchmarking Deep Research Agents in Misleading Evidence Environments

Deep research agents increasingly operate over the open web, where relevant records coexist with redundant summaries, outdated reports, and misleading documents. Existing evaluations offer limited insight into whether agents preserve sound evidential standards when an ordinary-looking false document is deliberately seeded into a searchable environment and offers a direct shortcut to a conflicting answer. We introduce DRNOISE, a 100-task benchmark for answer recovery under misleading evidence. Each task has a unique gold answer supported by two corroborating indirect record chains; the paired noisy condition adds one plausible document that states a conflicting answer directly. The benchmark spans ten families of evidence operations. Across agents with strong clean-task performance, this single intervention causes 66-88 percentage-point accuracy drops. Trace analyses identify verification inertia as the dominant failure mode: agents often retrieve truthful records but stop before completing and reconciling the evidence chain, instead deferring to the answer-like document. Generic verification prompts reduce but do not close this gap. The setting is especially relevant to open-web deployment, where plausible falsehoods arrive through ordinary-looking pages rather than explicit attacks. Reliable deep research therefore requires more than retrieval and citation; it requires active reconciliation of direct claims with record-level evidence.

Jun Nie, Zhiqin Yang, Zhenheng Tang +4

Hong Kong Baptist University · University of Science and Technology of China · The Hong Kong University of Science and Technology

Deep Research Falsehood Answer Self-Verification

Jul 15, 2026cs.CL

DS@GT ARC at LongEval: Citation Integrity and Factual Grounding in Scientific QA

This paper describes DS@GT ARC's submission to the CLEF 2026 LongEval Task 4 on Retrieval-Augmented Generation (RAG). In this submission, we examine a divergence between traditional natural language evaluation metrics and citation integrity as applied to RAG QA systems. We evaluate a corrective pipeline using Corrective RAG (CRAG) and CiteFix against baseline and frontier model benchmark RAG QA scores. While frontier models maximized answer relevance and fluency scores, our RAGAs LLM-as-judge diagnostics indicate that frontier models would correctly identify relevant documents without using their context in answer generation. Conversely, by filtering chunks pre-generation and enforcing strict entailment of generated claims to the cited material post-generation, our corrective pipeline marginally improved citation faithfulness and answer grounding. We propose that evaluation of trustworthy RAG QA requires metrics that reward strict answer grounding.

Brandon Michaels, Brendon Johnson

Georgia Institute of Technology, North Ave NW, Atlanta, GA 30332

Citations Explanation Faithfulness Answer Legal Natural Language Processing Benchmarks Text Corpora

Jul 15, 2026cs.LG

Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values

People use language models for practical questions whose answers are difficult to verify. We show that models exhibit covert value leakage: the information they provide is influenced by their own values, without this influence being disclosed to the user. In one of our evaluations, the user is considering investing in an AI company and wants to know how likely the AI bubble is to pop. Claude Opus 4.8 gives a lower probability when the company under consideration is Anthropic rather than OpenAI. Yet Claude mostly fails to disclose this influence to the user. Covert value leakage is a form of misalignment because it goes against the user's preferences and is likely to mislead them. To investigate this phenomenon, we introduce a suite of evaluations to quantify value leakage and whether models disclose it. We find that models are influenced by different types of values, including preferences for morally good outcomes, for the company that developed them, and for some human leisure activities over others. We often observe large differences among frontier models on the same evaluation. For example, on a Fermi-estimation task, Claude models falsely claim to give unbiased answers in their chain-of-thought, while Qwen models explain how their values bias their answers. Value leakage is a failure mode distinct from sycophancy and reward hacking, and current alignment training and evaluations do not adequately address it.

Jan Betley, Johannes Treutlein, Jan Dubiński +5

1Truthful AI · 2Warsaw University of Technology · 3NASK National Research Institute +2

Data Leakage Value Artificial Intelligence Systems Answer Chain-of-Thought Reasoning

Jul 14, 2026cs.CL

LLM Judges Can Be Too Generous When There Is No Reference Answer

LLM judges are increasingly being used to evaluate open-ended model responses, often in no-reference settings where a ground-truth answer is unavailable. However, can they reliably assess in such evaluation setups? We explore this question in this paper through a two stage pipeline with a) calibration experiments that assess the judge model's knowledge of the task it is evaluating, and b) sensitivity experiments that assess how the judge model's performance is impacted by the presence and positioning of the reference answer in the prompt. Across experiments covering three languages, we show that the judge models we evaluated tend to over-credit incorrect answers in the absence of a reference answer, and adding reference answer information to the prompt flips the judge model's correct/incorrect decisions by as much as 85% in some experimental settings. Comparison with a subset of human annotations shows that these reference-driven changes generally align with human judgments. Our results emphasize the need for calibrating the LLM judges with a sample with reference-aware evaluation before using them in reference-free setups reliably, and our methodology provides a blueprint for researchers and practitioners in doing such calibration of LLM judges for other tasks.

Chalamalasetti Kranti, Sowmya Vajjala

Computational Linguistics, Department of Linguistics, University of Potsdam, Germany · National Research Council, Ottawa, Canada

Large Language Model Judges Human Judgment Open-Ended Responses Answer

Jul 14, 2026cs.CL

The One-Word Census: Answer-Choice Conformity Across 44 Language Models

When a language model must pick one answer from a large space of equally valid options, which does it pick -- and how often is it the same answer every other model picks? Asked to "pick a word -- any word," 44 models chose "serendipity" 41% of the time. We characterize this convergence with a deliberately minimal instrument: 31 single-turn prompts, each naming a category with many valid one-word answers ("Name a tree."), asked four times per model with no system prompt. Analysis is exact-match on normalized tokens -- no embeddings, no judge -- at about a dollar per model. That models converge is well documented; our contribution is the instrument itself -- the One-Word Census -- and what it reveals about the structure of the convergence. We score each model by answer-choice surprisal: the average

-\log2

probability of its answers under the pooled answers of all other models, leave-one-out. Convergence is extreme -- in 7 of 31 categories one answer takes over 80% of all answers -- yet conformity varies more than fourfold across models, and the variation is structured. Persona- and community-tuned models are the most divergent; the newest mainline flagships are the most conformist, producing almost no answer no other model gave. Within four lineages (Claude, GPT, Qwen, Grok) conformity rises with each generation -- but reverses for the latest flagship Claude and GPT models, a possible early signal of repositioning at the top tier. Rankings are robust to roster composition (leave-one-family-out rho = 0.985). Against human category-production norms, the field is more concentrated than people in 18 of 20 shared categories. All prompts, transcripts, and code are public.

Tapan Parikh

Cornell Tech

Language Model Computes Answer Conformity Populations Convergence

Jul 14, 2026cs.CL

Policy-Conditioned Constrained Decoding for Column-Level Access Control in Text-to-SQL

Text-to-SQL is increasingly deployed across trust boundaries between data providers and users. Such deployment must balance three competing requirements: policy compliance, answer coverage, and bounded cost. Existing approaches typically decide refusal based on which columns a query mentions and enforce it stochastically. Whether a query is compliant, however, depends not only on which columns appear but on how they are used, and stochastic enforcement cannot deterministically rule out violations. We formalize this requirement as a column-use policy over semantic use: output, filter condition, and aggregation argument. We integrate the policy by aligning each role with grammar productions tracked by the decoder. The resulting system, PCC-SQL, applies a per-token logits mask that deterministically eliminates single-query column-use violations on the supported SQL fragment in a single decoding pass. Across three benchmarks and three open-source models, PCC-SQL achieves 0% Leakage Rate and Coverage up to 88.7% on Spider-CU, while staying within +10% tokens of direct prompting. We additionally assess semantic alignment with execution accuracy.

Ryoto Miyamoto, Xin Fan, Hayato Yamana

Waseda University

Text-To-Sql Access Control Data Leakage Queries Compliance Constrained Decoding+2

Jul 13, 2026cs.LG

PRISM Edit: One Vector for All Temporal Answers

Model editing keeps large language models (LLMs) up to date without retraining, but temporal facts expose a limitation of the prevailing locate-and-edit paradigm: an update is not always a replacement. When a fact changes, the new answer should become current while the old answer may remain correct in historical time contexts. Building on this insight, we use causal tracing to show that LLMs already support this distinction via a two-stage internal computation: early MLP layers retrieve a time-agnostic subject representation, and later layers modulate it with temporal context to yield the time-correct answer. Motivated by this finding, we introduce PRISM Edit, which optimizes a single polysemous representation across temporal contexts and leverages the model's inherent modulation pathway to route it to temporally correct predictions without requiring any architectural modification. We evaluate on TimeConflict, a newly introduced temporal editing benchmark, and on temporally augmented CounterFact. PRISM Edit improves multiple core metrics over the best baseline, most notably +23.3 Temporal Consistency (TC) and +33.7 Current Relative-time Score (CRS) on LLaMA-3, while being more than 2x faster. Code and data are publicly available at https://github.com/CheerCHuang/PRISM-Edit.

Chen Huang, Qi Zheng, Ruiqin Zheng +2

Tsinghua University · ByteDance

Multimodal Knowledge Editing Answer Prism Vectors Multilayer Perceptrons

Jul 12, 2026cs.AI

STEC: Evidence Compression for Deep Search in Open-domain Multi-Hop QA

In open-domain multi-hop question answering (QA), LLM-based search agents offer a promising approach to knowledge-intensive QA by combining retrieval with reasoning. Existing methods mainly improve open-domain multi-hop QA through reasoning paradigms, retrieval interaction, and search strategy optimization. However, using multiple search trajectories introduces a challenging final answer selection problem. Different trajectories may support different candidates, and the retrieved information can be heterogeneous, redundant, incomplete, or conflicting. Directly comparing raw trajectories exposes the verifier to noisy and unaligned content, while comparing answer strings ignores the evidence supporting each candidate, making reliable final selection difficult. To address this challenge, we propose STEC, an evidence compression framework for final answer selection in multi-hop QA. STEC selects the final answer from the existing candidate set through two mechanisms: (1) Answer-Level Evidence Compression, which groups trajectories by normalized answer identity and converts each answer group into a candidate-specific evidence representation; and (2) Evidence-Guided Answer Verification, which compares these representations and selects the final answer from the candidate set. The design shifts final selection from raw trajectory comparison to candidate-level evidence comparison. We evaluate STEC on four open-domain multi-hop QA benchmarks against representative baselines. Experimental results show that STEC performs best overall among the compared methods, and ablation results provide evidence that answer-level evidence compression contributes to final answer selection.

Xinkang Li, Rong Jiang, Xin Song +3

National University of Defense Technology, China · Harbin Institute of Technology, Shenzhen, China

Multi-Hop Reasoning Multimodal Query Answer

Jul 11, 2026cs.CV

What Does Your Short-Answer VQA Score Actually Measure? Evaluator-Dependent Instability in Multimodal Short-Answer Benchmarks

Short-answer VQA benchmarks conflate two distinct quantities: whether a model's answer is semantically correct, and whether that answer matches the surface form expected by the automatic evaluator. We study this conflation across six vision--language models and six benchmarks, using a human-validated semantic judge (97.6% precision) to audit over 37k official errors. A second text-only judge reproduces the same benchmark-level false-negative pattern, showing that the effect is not an artifact of a single audit model. On text-rich benchmarks, up to half of these errors are semantically acceptable answers penalized purely for surface-form mismatch. This instability is structured by answer type: extractive and multi-span answers are far more evaluator-sensitive than scalar answers. Benign prompt and context rewrites further destabilize official outcomes, flipping item-level correctness at substantial rates without changing the underlying task. A deterministic CPU-only contract repair confirms that the undercount is partially recoverable. These findings imply that official short-answer VQA scores should be accompanied by semantic audits and answer-type diagnostics to remain interpretable.

Guanhua Ye, Niu Jingbin, Yan Li +4

Beijing University of Posts and Telecommunications Beijing, China

Multiple-Choice Visual Question Answering Answer Large Language Model Judges Multimodal Benchmarks Scores Diagnosis

Jul 11, 2026cs.CL

Similar Accuracy, Unequal Evidence: Search APIs as Decision Surfaces for Tool-Using Agents

Search APIs expose ranked snippets, URLs, and metadata on which agents decide whether to answer, search again, or fetch pages. We evaluate these interfaces as decision surfaces on a fixed sample of 100 questions from the 254-question SealQA-Hard subset, using one frozen GPT-5.4 agent, a fixed orchestration harness, and a shared page-fetch backend across Brave, Tavily, and Firecrawl. A Kimi-K2.6 oracle labels visible URL-level evidence; a separate answer audit yields 25, 25, and 26 correct answers out of 100. These counts indicate similar observed accuracy but do not establish equivalence. Under the tested configurations, Brave exposes more pre-fetch support alongside a larger snippet surface; Tavily has a larger rank-1 share among trajectory-pooled supporting observations; and Firecrawl is associated with broader exploration. First-search query-level metrics distinguish support availability from ranking, while contradiction exposure complements contradiction-to-gold ratios. A retrospective oracle union covers 44/100 questions, versus 26/100 for the best individual provider: 18 percentage points of headroom. The observed evidence and action differences motivate evaluating search APIs jointly with agent policy and retrieval budget.

Sriram Selvam, Anneswa Ghosh

Retrievers Application Programming Interfaces Answer Decisions

Jul 11, 2026cs.AI

From ambiguous utterances to governed reuse classes: canonicalization, quotient invariance, and conditional decidability

Semantic caching defines answer reuse on embedding similarity: two utterances share a stored answer when a similarity score clears a threshold, with no notion of authorization, versioning, or of what makes two demands the same. This note changes the object on which reuse is defined: in a governed domain, reuse should operate on a mathematically characterized quotient of resolved conversational demands, not on a similarity heuristic. Three independently defined relations on resolved utterances -- reading identity, resolution identity, and reuse identity -- form a refinement chain, strict under realized nondegeneracy conditions checkable on deployment logs; the pipeline's outputs are invariant along the chain, and reuse identity is exactly the kernel of the resolution map into the governed answer partition, so the reuse quotient is the utterance-side object that partition induces, not a relabeling of it. Reuse identity licenses the governed query key and its certified answer space; reuse of a particular answer requires resolution identity or an applicability certificate. The supporting layer is stated at exactly the strength proved: exact-denotation normal forms; join aggregation as a design operator, with closure-stable cells characterizing no-escape; total computability of the full pipeline relative to an untrusted proposal layer; policy admissibility for arbitrary proposers -- and provably not factual grounding or intent fidelity; and elicitation terminating after finitely many informative replies, sound under target consistency.

Cosimo Spera, Ray Garcia

Minerva CQ (Bourbaki Intelligent Systems, Inc.), Los Gatos, CA, USA

Reuse Tractability Answer Utterances Similarity and Metrics Untrusted Content

Jul 9, 2026cs.CL

Two Axes of LLM Abstention: Answer Correctness and Question Answerability

A model should refuse two different things: answers it would get wrong, and questions it should not answer at all, such as unanswerable ones or ones resting on a false premise. The usual recipe thresholds a single confidence score, which cannot tell these apart. Across five instruction-tuned models from three families (2B to 14B), we find they are separate axes. Ordinary answer-confidence tracks whether an answer is right but is nearly blind to whether the question is answerable; a linear probe on hidden states does the reverse. The blind spot does not shrink with scale. It is worst on naturally occurring false-premise questions (CREPE). There, answer-confidence, P(IK), P(True), and even asking the model outright whether a premise is false all stay near chance, while a hidden-state probe reaches 0.69 to 0.77 AUROC: the model represents a problem it will not report. This turns out to be fixable. Instructing a model to check premises backfires, because it then disputes sound and false premises alike (57% false challenges), unable to tell them apart; routing the same instruction with the probe roughly triples challenge precision. We turn the two axes into a calibrated policy that answers only when an answerability score and a correctness score each clear a separately certifies behave differently: the unanswerable-answer rate is controllable at every scale, while the wrong-answer rate is capped by model accuracy, so the guarantee tightens as threshold policy certifies both budgets at 0.75 coverage of correct answers, against 0.31 for a single threshold; at 14B it is the only policy that certifies at all.

Benedikt J. Wagner

City St George’s, University of London

Answer Instruction-Tuned Models Question Decision Thresholds

Jul 9, 2026cs.AI

Answer Set Programming Energised! End-to-End Neurosymbolic Reasoning and Learning with ASP and Energy Based Models

We present a general neurosymbolic reasoning and learning methodology based on a modular integration of answer set programming with an energy based model substrate. Key contributions are: (1) supporting joint optimisation in the continuous latent space through explicit ASP-based declarative semantics fully incorporating background knowledge, constraints, non-monotonic inference; and (2) advancing recent works at the interface of answer sets, probabilistic logic, and answer set modulo theories by providing a generalised model and practical platform for ASP-centric robust, end-to-end training for applications in dynamic domains (e.g., involving perception and interaction). We provide a practical implementation, and demonstrate basic use and application (with MNIST), and evaluate with the visual question-answering benchmark Clevr and the multi-object tracking benchmark MOT.

Jakob Suchan, Julius Monsen, Salim Baloch +1

Constructor University Bremen, Bremen, Germany · CoDesign Lab > Cognitive Vision · Örebro University, Örebro, Sweden

Answer Set Programming Neuro-Symbolic Framework Efficient Latent Reasoning Transition Learning Answer Mnist

Jul 9, 2026cs.CL

What LLM Forecasters Know but Don't Say: Probing Internal Representations for Calibration and Faithfulness

Large language models fine-tuned for forecasting can be accurate yet poorly calibrated, and their chain-of-thought (CoT) reasoning may not faithfully reflect the evidence behind a forecast. We ask whether internal representations offer a more direct window into both. Working with Eternis-Forecaster 8B on OpenForesight, we train representation-pooling probes on intermediate activations and find they achieve substantially better calibration; a result that also holds for GLM-4.7-Flash and GLM-4.5-Air. We then assess CoT faithfulness through evidence ablation and diversionary injection: removing an influential source in the prompt often changes the model's forecast while leaving the reasoning trace untouched. The same probes function as lie detectors: their activations track behavioral shifts far better than the reasoning trace does, and they also predict the direction of change in 84% of cases, including when the CoT conceals the perturbation's influence. Finally, forced answering reveals that forecasts are largely fixed before reasoning begins: a single pre-reasoning pass recovers the committed answer and confidence, and routing questions by the spread of this pre-set answer distribution saves 30-47% of generated tokens, with no loss of accuracy. Together, these results establish probing internal representations as a practical tool for calibrating, auditing, and triaging language model forecasters and reasoning models more broadly.

Raphaël Sarfati, Pratyush Ranjan Tiwari, Siddharth Boppana +3

goodfire · eternis

Large Language Model Forecasting Data-Driven Forecasting Reasoning Skills Chain-of-Thought Reasoning Faithfulness Answer

Jul 7, 2026cs.LG

Final Checkpoints Are Not Enough: Analyzing Latent Reasoning Faithfulness Along Training Trajectories

Latent reasoning performs multi-step inference in continuous hidden states, promising more compact and efficient reasoning. However, these opaque states raise a question of faithfulness: whether the latent reasoning steps drive the final answer. Prior work studies this question at selected checkpoints and reports several unfaithful behaviors. This endpoint view leaves how evidence of faithfulness evolves during training unexamined. We track behavioral and activation-based evidence across training using verified counterfactual edits and interventions on the latent reasoning states. We find that high task accuracy can coexist with low counterfactual responsiveness: as accuracy improves, responsiveness can decline, and different latent reasoning approaches follow distinct trajectories. On ProsQA, output sensitivity to norm-noise replacement declines alongside counterfactual responsiveness, although the result depends on the replacement. Across separately trained binary-choice and open-ended GSM settings, intervention sensitivity follows opposite trajectories. These results show that evaluating only a final checkpoint can obscure both when counterfactual responsiveness changes and what the latent states contribute.

Hengyu Jin, Shu Yang, Di Wang

Tongji University · Provable Responsible AI and Data Analytics (PRADA) Lab · King Abdullah University of Science and Technology

Efficient Latent Reasoning Explanation Faithfulness Reasoning Traces Answer Intermediate Checkpoints

Jul 6, 2026cs.AI

Memory in the Loop: In-Process Retrieval as ExtendedWorking Memory for Language Agents

Language agents run a loop - observe, reason, act - but the memory they reason over sits outside it: a store queried at most once per turn. We study the regime where memory moves inside the loop, read and written on every step. The obstacle has always been latency: networked stores answer in tens to hundreds of milliseconds, and in-loop retrieval can inflate end-to-end latency by up to 83x when retrieval is expensive. Prior work manages that cost rather than questioning it: serving-layer scheduling hides it, "memory-first" designs ration retrieval to once per turn. We argue latency is a property of where the store lives, not the in-loop pattern: an in-process store answers in ~100us, three orders of magnitude below the network regime, and at that speed the per-step tax collapses. By the extended-mind thesis's parity principle, a store fast enough to be constantly and directly available becomes extended working memory, not a tool the agent merely consults. The premise is causal: holding a fixed per-turn memory-latency budget and varying only the store's answer speed, redundant actions rise monotonically with latency - 0.0 of 12 at in-process speed, 7.2 of 12 at a 110ms cloud round trip (gpt-5-nano, gpt-5-mini; exact permutation p=0.0079). We demonstrate the regime end-to-end: across four GPT-5-class models under a bounded window, recall improves from 0/5 to 3.6-4.8/5 with in-loop memory, store ops at p50 80-165us - though an instructed restate-every-reply baseline also solves it perfectly, at a token cost that grows with the working set. The store never lost a fact in any run (244 of 244 writes kept); every miss traces to the agent's read policy, not the store. Our measurements also relocate the bottleneck: the dominant per-step cost is embedding (~200-400ms over the network); pairing the in-process store with a small local embedder returns the complete operation to a measured ~40us.

Yusuf Khan, Carlo Lipizzi

Stevens

Working Memory Loop Answer

Jul 6, 2026cs.CL

Most LLM Conformity Needs No Speaker: Measuring the Speaker-Free Floor in Peer-Pressure Benchmarks

LLM conformity is often used to describe cases where a model changes a correct answer toward a peer or group response. We show that most of this apparent conformity survives even after the peer is removed. The reason is a confound: standard conformity prompts mix two cues at once, the presence of a speaker and the repeated wrong answer itself. Existing benchmarks vary these cues together, so they cannot tell how much of the revision actually depends on the speaker. We introduce a no-source condition: the same asserted answer with the explicit speaker removed. Across six open-weight LLMs and seven QA and reasoning datasets, this condition alone causes harmful revision in

66.5\%

of initially correct cases, compared with

10.3\%

under a plain re-ask. The effect also remains when the repeated answer is paraphrased and when answer options are hidden in an open-ended setting. Source framing mainly modulates this floor: expert-panel framing raises it, while minimal person labels do not reliably raise it. When models flip, they are usually confidently wrong, and simple recalibration does not recover the original answer. Source attribution still matters, but it should be measured as an increment above this speaker-free floor. The methodological lesson is that conformity benchmarks should first measure what remains after the speaker is removed; without this step, benchmarks may mistake repeated text for social influence.

Yibo Hu, Jiaming Qu

Illinois Institute of Technology · Amazon

Conformity Answer Speaker Social Influence Floor

Jul 6, 2026cs.CV

ChatImage: Navigating Long-Form LLM Answers through Interactive Images

Large Language Models (LLMs) can produce detailed answers to complex queries, but these answers are typically presented as dense linear text, which makes fine-grained inspection, navigation, and return visits difficult. We present ChatImage, a system that converts long-form LLM answers into interactive visual images. Given a textual answer, ChatImage first normalizes its content into structured visual modules, plans a visual layout, and renders a coherent image. It then applies a second grounding pass to the rendered image with vision grounding models such as LocateAnything and MiMo-Vision, with optional SAM-style mask refinement, to identify the visible regions that should support interaction. From these grounded regions, ChatImage overlays transparent clickable hotspots on the image. Each hotspot opens a detail panel and a region-scoped follow-up thread, allowing the user to inspect and query a specific part of the answer without re-reading the full response. Instead of treating planned coordinates as the final interaction geometry, ChatImage uses them as priors and grounds the interaction targets after rendering, which improves consistency between visual content and clickable regions. We release a reference implementation and introduce a 30-question benchmark covering infographic, map, and scene-based answer formats. Evaluation with configured external models reports interaction-loop completion, a strict visual-alignment gate, and a SAM-based mask-completeness diagnostic.

Wencan Jiang, Jiangning Zhang, Yong Liu

1Zhejiang University

Large Language Model Responses Answer Layouts

Jul 6, 2026cs.AI

Detecting Answer-Driven Reasoning in LLM-Based Educational Tutors via Truncated Chain-of-Thought Auditing

Large language model (LLM) tutors often produce fluent step-by-step explanations, but a correct and pedagogically formatted response does not guarantee that the answer was derived from the student-facing problem. In realistic tutoring systems, the model may also have access to teacher notes, answer keys, rubrics, or retrieved solution artifacts. We study whether such private answer information can make tutor explanations answer-driven: the final answer is behaviorally available before the written explanation has justified it. Using Truncated Reasoning AUC Evaluation (TRACE), which probes how early a chain-of-thought prefix can pass a verifier, we evaluate 1000 GSM8K test problems under three paired tutoring contexts: question-only, correct answer-key, and wrong answer-key. At fixed fractions of each generated explanation, we force the model to answer immediately and verify the response against the gold numeric answer. With Qwen2.5-3B-Instruct, answer-key access raises median TRACE AUC from 0.375 to 0.900 and makes the gold answer available at the first 10% prefix in 997 of 1000 cases. The effect remains strong on the 746 examples where both question-only and answer-key explanations end with the correct answer. These results support truncated CoT auditing as a lightweight process-level diagnostic for answer-driven reasoning in math tutoring explanations.

Bonan Shen, Dingyan Shang, Youting Wang +1

Independent Researcher United States · Northeastern University Boston, MA, USA · Syracuse University Syracuse, NY, USA

Tutors LLM Reasoning Strategies Answer Large Language Model Responses Chain-of-Thought Reasoning Roc-Auc

Jul 5, 2026cs.CL

Uncertainty-Aware Abstention in Large Language Models with Provable Alignment Guarantees

Large language models (LLMs) are increasingly deployed in question answering (QA) systems, yet they may generate hallucinated or misaligned responses without reliable confidence estimates. Uncertainty quantification (UQ) offers a natural basis for selective answering, where a system answers only when its prediction is deemed reliable and abstains otherwise. However, existing uncertainty scores for LLMs are often heuristic: a threshold chosen on such scores does not, by itself, provide statistical guarantees on the error rate among accepted answers. We propose CIC, a confidence-interval-based calibration framework that converts arbitrary uncertainty scores into risk-controlled selective answering rules. Given a held-out calibration set, CIC evaluates each generated response using an application-specific alignment criterion and associates it with an uncertainty score and a binary error label. For each candidate uncertainty threshold, CIC estimates the acceptance-conditioned error rate and constructs a high-probability upper confidence bound using either Hoeffding-style or Clopper-Pearson confidence intervals. It then selects the largest threshold whose upper bound is below a user-specified risk level

α

, thereby maximizing the answering rate subject to a finite-sample reliability constraint. Under exchangeability, CIC guarantees with probability at least

1-δ

that the selected threshold, if non-null, controls the error rate among accepted answers at level

α

. We evaluate CIC on both closed-ended and open-ended QA benchmarks across seven LLMs and multiple uncertainty estimators. Experimental results show that CIC consistently achieves valid risk control while retaining strong answering efficiency, providing a practical and statistically grounded mechanism for deploying LLMs in reliability-sensitive QA workflows.

Sijin Dong, Hiroyuki Shinnou

1Ibaraki University, Japan.

Large Language Model Uncertainty Uncertainty Score Abstention Answer Feature Alignment

Jul 2, 2026cs.CL

Improving LLMs via Validator-to-Generator Alignment

Large language models are inconsistent: varying prompts or including unrelated information can lead to unexpected changes in model outputs. The generator-validator (G-V) gap is one manifestation of this phenomenon, where LLMs generate responses that they then deem as invalid if re-queried to validate them. In this work, we introduce a new formulation of G-V consistency that involves a principled correction for utterance frequency. Specifically, generators often assign low likelihood to valid strings simply because those strings are a priori unlikely, which makes naive notions of G-V consistency unworkable. We show that under a natural model of rational agents answering questions with multiple answers, consistency of the validator with a frequency-corrected generator score emerges naturally. Our method, \emph{\FCPAname} (\FCPA), is a training objective implementing frequency-corrected G-V consistency for real-world LLMs. Our experimental results show that training with \FCPA{} substantially improves both G-V consistency and generator performance over prior methods, with gains of up to

+27

pp in Pearson correlation on IFEval and HumanEval, while preserving validator quality across all evaluated tasks.

Juan Diego Rodriguez, Jocelyn Zhang, Katrin Erk +1

Department of Computer Science, The University of Texas at Austin · Departments of Linguistics and Computer Science, University of Massachusetts Amherst · Department of Computer Science & Center for Data Science, New York University

Large Language Model Generation Answer

Jul 2, 2026cs.AI

A-TMA: Decoupling State-Aware Memory Failures in Long-Term Agent Memory

Long term memory lets LLM agents act as persistent assistants, but user facts change. A useful memory system must know what is true now, what used to be true, and what changed. We study \emph{ghost memory}, a state coordination failure in which old, current, and transition facts coexist in the memory bank, remain mixed during retrieval, and mislead the answer model. We argue that memory systems should be understood and optimized from three levels: bank maintenance, retrieval, and answer time resolution. We propose ATMA, a state aware overlay for existing memory systems. ATMA keeps superseded and transition records in the bank, builds evidence packets for the query's requested state view, and exposes current, historical, and transition labels to QA. We further call for decoupled evaluation of bank, retrieval, and answer level failures, since final QA accuracy can hide where ghost memory occurs. To make this failure measurable, we build LTP (LoCoMo Temporal Plus), a conflict heavy benchmark for ghost memory, and evaluate on LoCoMo for long conversation generalization. On LTP, Graphiti+ATMA improves conflict accuracy by 0.240 absolute over Graphiti. On LoCoMo, Graphiti+ATMA raises temporal F1 from 0.0295 to 0.1705. The gains are host dependent, but they indicate that explicit state roles can reduce memory failures hidden by final QA accuracy.

Zitong Shi, Yixuan Tang, Anthony Kum Hoe Tung

National University of Singapore

Long-Term Agent Memory Artificial Intelligence Assistants Answer

Jul 1, 2026cs.CL

What Survives Into Context: A Diagnostic for Budget-Constrained Multi-Hop RAG and When Submodular Evidence Packing Improves It

Retrieval-augmented generation (RAG) under a fixed reader-context budget forces a selection problem: of the evidence retrieved, only a fraction can be shown to the reader. We argue that document recall -- the standard retrieval metric -- is the wrong quantity to optimize in this regime, and we make two contributions. First, as a general contribution, we introduce answer-in-context, a diagnostic that measures whether a gold answer survives as a contiguous span in the packed reader context (not the retrieved set). It predicts answer F1 better than recall (r=0.39-0.55 vs. about 0.31), separates answer quality roughly five-fold (0.60 vs. 0.12 on HotpotQA), and carries information beyond retrieval: it adds Delta R squared=0.17 over recall and shows a 4.6x EM gap even among questions where all gold was retrieved. We also confirm it interventionally: on 2WikiMultiHopQA a packing change that raises coverage but not answer-in-context yields no accuracy gain. Second, as a conditional contribution, we cast reader-context construction as budgeted monotone submodular maximization and build a packer that jointly optimizes relevance, query coverage, representativeness, and diversity. On HotpotQA with a 160-token budget and a 3B reader it beats a strong focused heuristic, MMR, and naive packing -- by up to +5.1 F1 at equal-or-lower token cost, across three seeds. Crucially, we map the scope of this win honestly: it requires the conjunction of (i) multi-hop complementary structure, (ii) retrieval that surfaces the evidence, (iii) a binding but not extreme budget, and (iv) a reader weak enough that evidence density, not reading capacity, is the bottleneck. A quantization-controlled reader-scale ladder (3B to 7B to 14B) shows the edge over the heuristic is absorbed by 7B and significantly reverses by 14B, while the diagnostic explains every boundary with a single variable.

Ananto Nayan Bala

Ahsanullah University of Science and Technology

Answer Factual Recall Token Budget Allocation Submodular Functions Diagnosis

Jun 30, 2026cs.AI

Think in English, Answer in Korean: Efficient Adaptation of Multilingual Tool-Using Agents

We present LuckyStar 111B, a 111B-parameter hybrid reasoning model developed through a collaboration between Cohere and LG CNS for Korean-English enterprise agents under practical memory and serving constraints. The model trains from Cohere's fully post-trained Command A model rather than a new pretraining run, and uses preamble conditioning to switch between concise non-reasoning behavior and longer tool-oriented reasoning. We study four choices for scaling tool-using agents efficiently: multilingual supervised fine-tuning, reinforcement learning with verifiable rewards for multi-step tool-use tasks, language-consistency rewards for Korean user-facing responses, and 4-bit quantization for single-GPU serving. The adapted model improves mathematical reasoning, function calling, and agentic natural-language-to-SQL (NL2SQL) performance while preserving general Korean and English instruction-following quality. These results provide a practical recipe and failure-mode analysis for adapting post-trained multilingual models to verifiable agentic workflows under memory-constrained deployment.

Utsav Garg, Sungjin Hong, Jason Jung +6

1Cohere

Multilingual Agents Multilingual Language Models Answer Text-To-Sql

Jun 29, 2026cs.NI

Budget-Adaptive Routing: Skipping the Weak When the Strong Answers Anyway

Edge-cloud inference collaborations are often designed with a routing estimator that decides whether to offload each frame from weak models at the edge to stronger models in the cloud. Existing systems place the routing estimator after the weak detector, so the weak forward pass still runs even on frames that are later offloaded. In this paper, we argue that this weak-conditioned design can be suboptimal when the offload budget varies. First, we present a competitive weak-skipping estimator (0.153 GFLOPs, about 29x lighter than the weak detector at 4.49 GFLOPs) that extracts routing signal from raw pixels, outperforming the common after-weak placement weak-conditioned baselines. Second, we show that neither weak-skipping nor weak-conditioned placement dominates across the full operating curve, and we propose budget-adaptive routing, which selects between them by offload budget via two offline-tuned thresholds. On PASCAL VOC, our budget-adaptive router traces the upper accuracy envelope of both fixed placements across the operating range. Our method reduces per-frame latency by up to 19.1 ms (about 30% lower at rho = 0.9). Besides outperforming SOTA methods, it is surprisingly stronger than the strong model (+1.7 pp over the strong model's peak mAP) at some operating points with far less compute. Artifacts are available at https://github.com/ViGeng/bgt-ada

Wei Geng, Nitinder Mohan, Jörg Ott

Technical University of Munich Munich, Germany · TU Delft Delft, Netherlands

Jun 28, 2026cs.CL

Travel-Oriented Reasoning Large Language Model via Domain-Specific Knowledge Graphs

Large language models (LLMs) demonstrate broad reasoning abilities but struggle with accuracy and reliability in specialized domains such as travel, where reasoning depends on precise definitions, rules, and expert-defined conceptual frameworks, and where confident but unfounded outputs arise from a reasoning failure in which the model has not internalized the underlying domain graph rather than from missing domain knowledge alone. We propose a modular pipeline for building a travel-domain reasoning LLM grounded in an expert-designed knowledge graph (KG). Our pipeline integrates a travel KG that encodes domain entities and their relationships, a bottom-up construction procedure that walks the KG to produce multi-hop question answer (QA) pairs, a supervised fine-tuning stage that embeds the domain knowledge into a reasoning-capable LLM using the generated QA pairs as auditable reasoning traces, and a travel-domain benchmark dataset that measures the fine-tuned model's accuracy and calibration. We evaluate our approach using Qwen3-4B with LoRA adaptation. Our reasoning model achieves an

82.4\%

exact match on the benchmark. This performance significantly outperforms the pretrained Qwen3-4B baseline at

22.4\%

. A calibration analysis decomposes the residual

17.57\%

of errors into two distinct failure modes: an over-confident multi-label decoder that predicts both correct answers plus one spurious option on most dual-answer mistakes, and a smaller reasoning failure on single-answer questions where the supporting facts are present in the KG but the model fails to reconstruct the correct multi-hop path. This split confirms that explicit KG-grounded reasoning substantially improves the accuracy and uncertainty interpretation of LLMs in specialized domains, and isolates per-option calibration and trace-length-aware decoding as the next axes of improvement.

Vignesh Ram Nithin Kappagantula, Shayan Hassantabar, Samuel Simpson +1

Expedia Group Seattle, Washington, USA

LLM Reasoning Strategies Journey Reasoning Skills Knowledge Graphs Answer

Jun 25, 2026stat.ML

When are likely answers right? On Sequence Probability and Correctness in LLMs

Many decoding methods for large language models can be understood as shifting probability mass toward outputs that are more likely under the model, either locally at the token level or globally at the sequence level. Therefore, their success depends on a fundamental question: when does sequence probability, that is, the conditional probability of a continuation given a prompt, actually align with correctness? In this paper, we set out to quantify this relationship across decoding methods, models, and benchmarks at four levels: across decoding methods, across hyperparameters within a method, across prompt-answer pairs within a dataset, and across repeated responses to the same prompt. We find that higher sequence probability is often predictive of correctness across prompt-answer pairs within a fixed dataset. However, this relationship does not generally transfer to decoding decisions: increasing sequence probability by changing hyperparameters or methods does not reliably improve accuracy. Further, sequence probability is not a good indicator of correctness for responses to the same prompt. These findings clarify when decoding can and cannot be expected to improve correctness, and provide practical guidance for decoding, self-consistency, and verifier-free self-improvement.

Johannes Zenn, Jonas Geiping

Max Planck Institute for Intelligent Systems · ELLIS Institute Tübingen · AI Center Tübingen +2

Correctness Self-Consistency Answer Probability

Jun 24, 2026cs.CL

ProvenAI: Provenance-Native Traces of Evidence in Generated Answers

Retrieval-augmented systems routinely present citations alongside generated answers, yet a citation does not confirm that the corresponding source meaningfully shaped the output. This paper introduces ProvenAI, a framework that decomposes transparency in multi-hop question answering into three independently measurable layers: answer correctness, citation fidelity against benchmark supporting evidence, and per-document influence under leave-one-resource-out intervention. Targeting the HotpotQA distractor benchmark through a seven-stage pipeline covering data normalisation, retrieval indexing, citation-aware answer generation, attribution auditing, ablation-based influence estimation, batch evaluation, and interactive inspection, ProvenAI evaluates 7,405 validation examples drawn from a canonical corpus of 509,300 passages. The system achieves 53.53% answer accuracy alongside a mean citation-fidelity score of 71.55%, and a worked example surfaces what we call the citation-influence gap: a clean citation audit co-occurring with a profile in which one cited source registers only weak influence while seven uncited sources demonstrably shift the output. We formalise the relationship between the implemented surface proxy and a token-level KL-divergence target through a stated faithfulness condition, ground the framework in causal-mediation analysis and database-provenance theory, and discuss how the three measurement layers compose with cryptographic provenance architectures emerging in autonomous scientific discovery. ProvenAI establishes that meaningful transparency in retrieval-grounded QA requires traceable links across retrieved, cited, and behaviourally influential evidence as three distinct, independently measured layers.

Mohammad Faizan, Dalal Alharthi

College of Information Science University of Arizona

Citations Traceability Answer Transparency Traces

Jun 23, 2026cs.CV

video-SALMONN-R $^3$ : Learning to ReWatch, ReAsk, and ReAnswer for Efficient Video Understanding

Video large language models (LLMs) are often constrained by computation and memory budgets, leading them to use reduced frame rates and spatial resolutions, which may cause them to miss critical information for question answering (QA). A practical and efficient solution is a two-stage paradigm: first perform coarse video understanding to localize relevant segments, and then re-watch these segments at higher temporal or spatial fidelity. In this paper, we present video-SALMONN-R

^3

, the first end-to-end video-LLM that enables re-watch through reinforcement learning without relying on chain-of-thought (CoT) cold-start. This design removes the need for costly CoT data annotations and avoids CoT-based supervised fine-tuning (SFT), which can otherwise degrade the pretrained video understanding abilities. To address the mismatch between the reasoning-first behavior induced by re-watch and the answer-first tendency of pretrained video-LLMs, we propose a re-answer strategy, in which the model first produces a direct answer in the first watch and then refines it after re-watching. Finally, to improve question adherence during re-watching, we propose a re-ask mechanism that re-injects the query when revisiting localized segments. Experimental results show that video-SALMONN-R

^3

consistently outperforms both the base model and the QA-SFT baseline, while surpassing prior re-watch-based approaches with significantly lower computational cost. Code, models, and data will be publicly released upon acceptance.

Yixuan Li, Guangzhi Sun, Yudong Yang +1

Tsinghua University · University of Cambridge

Video Understanding Answer

Jun 22, 2026cs.CY

The Correct Answer Trap: Pedagogically-Grounded Detection and Feedback for Hidden Misconceptions

Automated feedback systems that rely on answer correctness will reinforce, rather than address, misconceptions when students reach the correct answer through flawed reasoning. We investigate automatic detection of these hidden misconceptions using 20,964 real student responses from the Eedi mathematics platform. Fine-tuned classifiers detect only 57% of these hidden misconceptions, and standard ML interventions do not improve on this. An open-weight reasoning model detects 84%, but at realistic prevalence, false alarms outnumber genuine detections roughly 8 to 1. We present a graduated assessment rubric that separates answer correctness from method validity, and propose a detect-verify-escalate pipeline that routes uncertain cases to diagnostic follow-up questions rather than directly to teachers. Two deployment modes adapt the pipeline: a teacher dashboard where the system filters a review queue, and an autonomous tutor where flags trigger low-cost formative follow-up.

Moiz Imran, Sahan Bulathwela

Department of Computer Science, University College London, The United Kingdom · Centre for Artificial Intelligence, University College London, The United Kingdom

Student Misconceptions Tutors False-Negative Rate Answer Feedback

Jun 22, 2026cs.LG

Position: Correct Answer, Wrong Mechanism -- When AI Scientists Defend General Claims Their Own Data Contradicts

AI scientist systems are described as tools, coauthors, or founders, but we evaluate them as if only the final answer matters. This position paper argues that outcome-only evaluation is insufficient, and that task outcome, mechanism fidelity, and epistemic honesty must be measured separately. Our evidence comes from 28 episodes of a coding agent attempting to rediscover a known particle identification observable in a Geant4 simulation, including an 8-episode probe across two additional frontier models. In 4/20 primary-model and 3/8 cross-model episodes, agents reach right-looking results through incorrect reasoning that breaks when conditions change, which we call Correct Answer, Wrong Mechanism (CAWM). Honesty and mechanism fidelity dissociate within a single agent trajectory. When given a partially misleading prior, all five agents reject the false component on evidence, yet one defends its chosen observable with physics inconsistent with its own data. In the simulation-based discovery setting studied here, coding agents prove reliable tools but unreliable scientific co-authors for open-ended claim-making, where co-author trust requires mechanism-fidelity verification they do not reliably self-apply. The failure is detectable, and we propose a lightweight test. A one-step regime-shift check needs only the agent's claim and flags the over-generalized cases. A companion recomputation flags the remaining cases when the correct observable is known. Together, these checks flag every CAWM case in this study.

Steven Young Eulig

Department of Physics and Laboratory for Particle Physics and Cosmology (LPPC), Harvard University, Cambridge, MA 02138, USA.

Artificial Intelligence Scientists Correct Answer

Jun 22, 2026cs.CV

Chains That See, Answers That Don't: A Multi-Aspect Evaluation Recipe for Forced Chain-of-Thought on Video-MME

Forced chain-of-thought (CoT) is widely assumed to make vision-language models more reliable on video question answering. We propose a small three-probe evaluation recipe to test that assumption: paired accuracy across direct, CoT, answer-first, and no-video conditions; a counterfactual video-swap diagnostic over the CoT chains; and a four-rung visual-degradation ladder. Each probe is reported under both a strict and a permissive regex scorer, with multiplicity correction over a manuscript-declared primary family. Applied to Qwen2.5-VL on Video-MME subsets, the recipe returns a two-part finding. The CoT chains are strongly video-conditioned: swapping the input video collapses chain overlap and flips most final letters, the opposite of what a "boilerplate-chain" null would predict. Yet on the same data, forced CoT does not improve MCQ accuracy, and on the smaller 7B model it produces a small but statistically supported drop under a post-hoc primary scorer choice. We do not claim this generalizes beyond the Qwen2.5-VL / Video-MME instantiation; the raw responses and a single recomputation script will be released with the supplementary material so every number can be re-derived.

Zhichao Fan, Yanhang Li, Zexin Zhuang

University of Illinois Urbana-Champaign Urbana, Illinois, USA · Northeastern University Boston, Massachusetts, USA · Southern Methodist University Dallas, Texas, USA

Long Video Question Answering Chain-of-Thought Reasoning Multiple-Choice Questions Answer

Jun 20, 2026cs.CL

OpenBioRQ: Unsolved Biomedical Research Questions for Agents

A working citation looks like proof -- but the fact that a link resolves does not mean the cited paper supports the claim. I find that current agentic models rarely fabricate citations (over

99\%

resolve), yet roughly

15.9\%

link to the wrong paper. Existing benchmarks miss this failure mode: when a question has a fixed answer key, a model can reproduce the expected source from that key rather than independently verifying that the source supports the claim. I introduce \textbf{\openbiorq{}}, a retrieval-grounded agentic benchmark of

12{,}553

unsolved biomedical research questions across

12

domains that treats open questions as a faithfulness-and-abstention probe. To my knowledge, this is the first biomedical benchmark to combine an agentic setting -- where the model must issue multiple tool calls -- with unsolved questions that have no answer key. Openness is verified against real follow-up evidence rather than a model's parametric knowledge. Difficulty is empirical: I anchor it on questions that three open-weight reference models fail to answer, rather than on subjective hardness labels. On this hardest subset, held-out models from the same lineage as the difficulty anchors solve only ~17%, while three independent frontier agents (Gemini-3-Pro, Opus-4.7, GPT-5.5) span a wide 29-60% range. The benchmark is thus hard, non-saturating (the best agent still leaves ~33-40% unsolved), and discriminating across capability tiers. Beyond difficulty, I observe agentic collapse on the hardest questions, where agents stop using their tools. For the most collapse-prone model, blocking tool access entirely barely changes its score -- so tools stop paying off exactly where they are needed most. A frozen per-question checklist raises inter-judge agreement from Spearman 0.35 to 0.82.

Minbyul Jeong

Upstage AI

Biomedical Text Agentic Benchmarks Biomedical Citations Question Answer+1

Jun 20, 2026cs.LG

Ranking-and-Selection with Multiple Correct Answers and Non-Answerable Estimates

We study fixed-precision ranking-and-selection in structured settings where the answer may be non-unique and where noisy estimates may temporarily admit no valid answer at all. This phenomenon arises naturally in problems such as multi-fidelity ranking-and-selection and identifying a Condorcet winner from pairwise comparisons. To address this, we propose a unified framework based on answer-wise acceptance sets, restricted generalized likelihood ratio stopping, and an answer-pitfall decomposition that yields a max-max-min characteristic value and a common sampling principle. We introduce ENDS, a general procedure that combines estimation, nomination, pitfall detection, and cost-aware information-directed selection. We instantiate ENDS for various problems by deriving explicit formulas. Extensive numerical experiments show that this unified recipe performs well across a broad range of pure-exploration problems and offers a practical framework and proof-of-concept algorithmic recipe.

Qiaoqiao Wang, Wei You

Dept. of Industrial Engineering and Decision Analytics, The Hong Kong University of Science and Technology, Hong Kong, China

Selection Bias Ranking Finite-Sample Answer Stochastic Exploration Acceptance+1

Jun 19, 2026cs.AI

Answer Engineering: Local Trajectory Editing for Protocol-Constrained Decision Making in Large Language Models

Large language models can produce confident but protocol-invalid answers in domains where procedural compliance is critical. This paper presents Answer Engineering, a deterministic runtime and authoring layer that applies localized rule-guided interventions to the visible reasoning trajectory during standard autoregressive generation, without retraining, modifying model weights, or performing global search. The method is evaluated on a controlled clinical benchmark for sudden sensorineural hearing loss (SSNHL), where correct management depends on protocol-consistent interpretation of symptom timing, Weber/Rinne tuning-fork findings, and otoscopic findings. In the benchmark, step-by-step reasoning shifted rather than eliminated errors: compliant outcomes for SSNHL decreased from 54.5% under unguided generation to 25.1%, while acceptance on the conductive contrast condition increased from 1.6% to 58.9%. Local trajectory editing increased SSNHL compliance to 83.5% and conductive-case adherence to 77.9%, raising balanced accuracy from 42.0% under reasoning-only generation to 80.7%. The results support a systems-level view in which protocol adherence can be improved through auditable runtime control of reasoning trajectories, while also identifying limitations caused by rule coverage, trigger reliability, and persistent diagnosis-first generation dynamics.

Victor Lavrenko, Anastasiia Molodnitskaia

PeaceTech VC, Israel · Rambam Health Care Campus, Israel

Answer Engineering

Jun 18, 2026cs.AI

Optimal Scheduling in a Question-Answering Forum of Knowledge Workers

As individuals turn to the Internet to find answers to questions they may have, several Question Answering (QA) forums have evolved, where users knowledgeable in certain topics can contribute their expertise to answering these requests for information. While these are currently volunteer based, we consider a future version employing knowledge workers who are experts in certain topics. In such a system, the request-answer processes forming the queuing system may utilize schedulers that assign requests in different topics to the experts in the forum, who may be able to answer them according to their expertise levels in different topics. With this model, we calculate the capacity of the system for handling the requests while keeping the system stable, and design schedulers that achieve capacity. We also investigate how collaboration between experts in answering requests can potentially increase capacity.

Rohit Negi, Mustafa Yilmaz

Carnegie Mellon University

Schedulers Crowdsourced Framework Answer

Jun 17, 2026cs.LG

Does VLA Even Know the Basics? Measuring Commonsense and World Knowledge Retention in Vision-Language-Action Models

Embodied Vision-Language-Action (VLA) models are typically obtained by fine-tuning powerful pretrained VLMs on robotics data, yet it is unclear how much commonsense and factual knowledge they retain after adaptation. Failures on knowledge-sensitive tasks are ambiguous, conflating missing knowledge with poor generalization of low-level control. We introduce Act2Answer, a lightweight protocol that adapts VLM knowledge benchmarks to VLA evaluation by requiring agents to answer through action. Each question becomes a short tabletop episode where the agent performs a single object-placement action to select among candidate answers, yielding an action-grounded success rate with reduced control confounds. We curate a test suite of such environments across diverse commonsense and world-knowledge categories and introduce layerwise intent probing to localize answer-relevant information across the VLM backbone and action head. In a large-scale study of 7 VLA models and 9 VLM baselines, we systematically rank models across categories, finding that VLAs show solid performance on simple concepts while exhibiting larger gaps on richer semantic categories relative to their source VLMs, that VQA co-training is associated with better knowledge retention, and that answer-relevant signals peak in middle VLA layers but attenuate in upper layers. Act2Answer is available at https://tttonyalpha.github.io/act2answer/.

Nikita Kachaev, Andrey Moskalenko, Matvey Skripkin +10

1CogAI Lab · 2FusionBrain Lab · 3IAI MSU +8

Vision-Language Foundation Models Task Success Rate Answer

Jun 16, 2026cs.CV

See First, Answer Later: Visual Evidence Pre-Alignment via Sufficiency-Driven RL

Multimodal large language models (MLLMs) integrate strong text reasoning with visual inputs, yet their responses can be inconsistent with the underlying images, indicating ineffective utilization of visual evidence during inference. The prevailing training paradigm relies on large-scale caption-based pretraining for general alignment, followed by supervised fine-tuning and reinforcement learning to enable instruction following and complex reasoning. However, such pretraining provides only weak visual grounding: short, coarse captions bias models toward salient objects while neglecting fine-grained visual evidence. In this paper, we introduce Visual Evidence Pre-Alignment (VEPA), an intermediate stage between pretraining and post-training that explores a novel sufficiency-driven objective with Group Relative Policy Optimization (GRPO) to optimize question-conditioned visual evidence descriptions. Extensive experiments across diverse benchmarks show that our VEPA consistently enhances performance on visually demanding evaluations and complements standard supervised post-training. Further analyses show that the income stems from strengthened, transferable visual grounding, rather than from additional task-specific training.

Yilian Liu, Sicong Leng, Guoshun Nan +7

Beijing University of Posts and Telecommunications, China · Nanyang Technological University, Singapore · China Telecom, China

Visual Evidence Multimodal Large Language Models Offline Reinforcement Learning Answer

Jun 14, 2026cs.CL

Who Flips? Self- and Cross-Model Counterarguments Reveal Answer Instability in LLMs

Standard accuracy benchmarks evaluate whether large language models (LLMs) reach correct answers. However, they do not test whether models maintain that answer when challenged by a plausible counter-argument. We introduce a controlled protocol for evaluating answer stability: after a model answers a multiple-choice question correctly, we challenge the model's answer with a coherent argument for an incorrect option and measure whether the model flips. The setup a) isolates argumentative content from overt social pressure and b) varies argument length, self-attribution, and cross-model source. Across seven frontier models and 57 MMLU subjects, flip rates range from 17.5% to 97.3%, revealing large differences in stability that are not captured by accuracy metrics alone. We find that self-attribution consistently increases flip rates (mean 7.1pp, up to 18.7pp). Furthermore, pooling wrong-answer arguments across models and selecting the most effective one per question yields stronger adversarial challenges than relying on any single source model. From this cross-model pool, we construct MaxFlip, a curated benchmark that amplifies answer flips by up to 23.6pp over self-generated challenges. We release the protocol, challenge records, and MaxFlip to support stability evaluation alongside standard accuracy benchmarks. Materials are available at https://github.com/nafisenik/WhoFlips, https://hf.co/datasets/nafisehNik/WhoFlips.

Nafiseh Nikeghbal, Amir Hossein Kargaran, Shaghayegh Kolli +1

1Technical University of Munich · Munich Center for Machine Learning · 1Technical University of Munich 2 Munich Center for Machine Learning +2

Flips Answer Instability Adversarial Examples

Jun 14, 2026cs.CL

Formalize Once, Edit the Rest: Efficient Lean-Based Answer Selection for Math Reasoning

With large language models (LLMs) increasingly applied to mathematical reasoning, formal proof assistants such as Lean can be leveraged to verify reasoning outputs with machine-checkable rigor, enabling use cases such as answer selection in test-time scaling with K sampled candidate answers. However, employing Lean requires that LLM outputs, originally in natural language, first be formalized. Existing Lean-based answer-selection work uses an autoformalization model to generate a formal statement in Lean for each candidate answer independently, incurring a significant computational cost. We propose BASE, a base-and-edit pipeline that formalizes a single base candidate per problem and derives the remaining K-1 statements by editing the answer expression in place. To facilitate this, we train a rewriter model LEANSCRIBE to localize the answer in the base formalization and generate a reusable edit function for the other K-1 candidates. BASE simultaneously improves selection accuracy and reduces formalization cost - a Pareto improvement that holds on all 12 (dataset, solver) configurations across four benchmarks and three solvers, cutting autoformalizer calls by about 5x at K=8, with the reduction expected to become larger as K grows. Code is available at https://github.com/ucr-rai/base-and-edit.

Ji Feng, Zhouxing Shi

University of California, Riverside

Autoformalization Formalization Answer Rewrite

Jun 11, 2026cs.LG

Beyond the Commitment Boundary: Probing Epiphenomenal Chain-of-Thought in Large Reasoning Models

Chain-of-thought (CoT) reasoning is the dominant paradigm for inference-time scaling in language models, yet the causal influence of individual steps on the final answer remains poorly understood. In this work, we use answer logits at the end of each reasoning step to estimate each step's causal importance to the final answer and intermediate guesses, shedding light on the answer formation process of several reasoning model families. Across diverse tasks, we find that reasoning typically crosses a commitment boundary, a sharp transition from transient intermediate guesses to a stable, high-confidence answer. This transition often happens in a single step, well before the model's reasoning block ends, and is followed by epiphenomenal CoT steps that leave the final answer probability unaltered. Using attention probes, we show that answer-formation stages can be linearly decoded from the activations of intermediate reasoning steps with high accuracy, showing robust generalization to unseen reasoning tasks. We leverage this property for early-exiting reasoning blocks at the commitment boundary location, reducing the length of CoTs up to 55% with negligible impact on model performance.

Daniel Scalena, Sara Candussio, Luca Bortolussi +3

CLCG, University of Groningen · University of Milano-Bicocca · University of Trieste +1

Chain-of-Thought Reasoning Inference-Time Commitment Answer

Jun 10, 2026cs.CL

Semantic Grading of Written Answers in Low-Resource Language Bangla Using a Fine-Tuned Lightweight Language Model

Bangla is among the world's most widely spoken languages, yet it remains underserved in educational NLP research. In many remote and rural regions, access to qualified subject teachers is limited, and written answers are consequently graded largely by hand, restricting timely and consistent feedback. Automatic assessment is challenging because semantically correct responses can vary substantially in surface form. We present a bilingual (Bangla-English) evaluation system designed for low-resource educational settings that prioritizes semantic correctness over lexical overlap. Our approach fine-tunes a lightweight language model to grade each response using the question, reference answer, and student answer, producing a numeric score and concise, context-grounded feedback suitable for classroom deployment. We also construct a synthetic bilingual dataset to enable controlled training and evaluation. Across proprietary and open-source LLMs evaluated under a unified protocol, our QLoRA-tuned Qwen3-8B confirms consistent improvement by producing the most leakage-resistant feedback (RoRa = 0.819) in synthetic evaluation and the strongest agreement with human scores (rho = 0.936, MAE = 0.725) in a dedicated human study.

Meherun Farzana, Aniket Joarder, Mahmudul Hasan +1

Computer Science and Engineering, University of Dhaka

Bangla Grading Natural Language Processing Large Language Model Fine-Tuning Large Language Model Training Answer+1

Jun 9, 2026cs.CV

Towards Fully Automated Exam Grading: Fairness-Aware Recognition of Handwritten Answers with Foundation Models

Correcting handwritten exams by hand is time-consuming and error-prone, particularly for large cohorts, while fully digital exams tend to force a didactic narrowing towards closed question formats. A practical middle ground keeps paper-based, problem-oriented tasks but records the assessment-relevant answers as single capital letters in a table that a machine can read. The open question is whether this reading can be made accurate and, above all, fair enough for unsupervised grading. Earlier automated approaches reached only about 88%--91% recognition -- too low -- and failed on the cases that matter most: answers placed outside the cell, crossed out, or written in cursive. We show that general-purpose vision-language foundation models (VLMs), which interpret the page rather than match pixel templates, close this gap. On a benchmark of 61 anonymised exams (3141 answer positions) the best model reaches 98.4% accuracy, well above the previous baseline. Crucially, we centre the evaluation on fairness: we distinguish false negatives (a correct answer marked wrong, which disadvantages the student) from false positives, and a lightweight prompt that supplies the reference solution as context lowers the false-negative rate to 0.58%. Under an exemplary grading scheme only three of the 61 exams would be graded worse, all caught by a student self-review step. Fully automated, fairness-aware exam grading at scale is therefore defensible; we release the anonymised benchmark to support reproducibility.

Hartwig Grabowski

Institute for Machine Learning and Analytics (IMLA), Offenburg University, Offenburg, Germany.

Grading Handwriting Assessment False-Negative Rate Algorithmic Fairness Answer+2

Jun 9, 2026cs.LG

Divide and Cooperate: Role-Decomposed Multi-Agent LLM Training with Cross-Agent Learning Signals

Modern language agents which perform multi-step reasoning have shown strong performance in knowledge-intensive question answering. However, existing approaches typically couple evidence acquisition and answer generation within a single policy. This forces a single model to play multiple potentially conflicting roles, inducing a combinatorial explosion in the policy space and hindering efficient exploration. It also introduces a credit assignment problem during training: a search action that retrieves sufficient evidence may still be penalized when generation fails, and vice versa. We propose DAC (Divide and Cooperate), a role-decomposed multi-agent training framework that divides agentic search into two cooperative subtasks, each handled by a dedicated agent trained with role-specific learning signals. The generator serves a dual role as both an answer producer and an evidence sufficiency verifier, abstaining when retrieved evidence is insufficient. This abstention signal is incorporated into the search agent's reward, providing structured cross-agent learning signals that improve credit assignment. Conversely, the searcher exposes the generator to diverse and challenging evidence environments by hard-positive evidence augmentation, improving its robustness. Experiments on general and multi-hop QA benchmarks show that DAC, implemented via parameter-efficient LoRA modules over a shared backbone, achieves strong performance against prior baselines that rely on full fine-tuning of monolithic models.

Jaewan Park, Solbee Cho, Jay-Yoon Lee

Seoul National University

Mixture-Of-Agents Multi-Agent Large Language Model Systems Agentic Search Cooperation Answer

Jun 6, 2026cs.AI

When No Answer Is Correct: Diagnosing Absent Answer Detection for MLLMs in Video Understanding

Multimodal large language models (MLLMs) have made substantial advancements in video understanding, yet the reliability of their responses remains underexplored. This work presents a diagnostic study of absent answer detection for MLLMs in video understanding, where the correct answer is deliberately excluded from the candidate set and a reliable model is expected to recognize that no valid option exists. We evaluate the absent answer detection behavior under three settings: multiple-choice questions augmented with an ``None of the Above'' option, open-ended generation with a detection instruction, and standard evaluation without any guidance. Across a diverse set of models and benchmarks, we find that MLLMs overwhelmingly select plausible distractors rather than detecting the absent answer. This failure is more pronounced in temporal reasoning tasks and worsens with denser frame sampling. We further explore chain-of-thought prompting as a mitigation strategy and find that while it substantially improves detection rates, performance remains unsatisfactory, suggesting that prompting-based strategies alone are insufficient to fully address this limitation. These findings expose a systematic failure in absent answer detection and highlight the need for explicit detection mechanisms in multimodal systems.

Yiheng Wang, Yueqian Lin, Lichen Zhu +3

Duke University, Durham, North Carolina, USA

Video Multimodal Large Language Models Multimodal Large Language Models Video Understanding Multiple-Choice Questions Answer Temporal Reasoning

Jun 5, 2026cs.CV

VisualFLIP: Do Predictions Depend on Task-Critical Visual Evidence in Multimodal Reasoning?

When a multimodal large language model answers a visual reasoning question correctly, is the prediction actually supported by the task-critical visual evidence? Correct answers can coexist with flawed reasoning, making accuracy alone an incomplete test of grounding. We introduce VisualFLIP, a paired benchmark with 1,374 images arranged as same-question perturbation pairs across cardinality, attribute, spatial, and logic tasks. Each pair keeps the question fixed but minimally changes the evidence so the gold answer deterministically flips. We evaluate 24 MLLMs with pair accuracy, which requires solving both sides of a pair, and Collapse Rate (CR), which measures how often a model that solves at least one side repeats the same non-empty answer for both images. Together, these metrics show that paired correctness and evidence dependence are related but distinct: capable models can still fail to update after task-critical visual changes, and collapse becomes more severe for some models when the edited image follows an earlier answer in a sequential setting. Further details are available on our project page: https://didizhu-judy.github.io/VisualFLIP/

Didi Zhu, Changrui Chen, Stefanos Zafeiriou +1

Imperial College London

Multimodal Reasoning Benchmarks Multimodal Reasoning Visual Evidence Multimodal Large Language Models Visual Reasoning Flips+1

Jun 4, 2026cs.CL

Diagnosing Evidence Utilization in Long-Context and Retrieval-Augmented Language Models under Matched Evidence Conditions

Final-answer accuracy, retrieval recall, and citation overlap do not reveal how much answer advantage a long-context or retrieval-augmented language model actually recovers from supplied evidence. A model may answer from parametric priors, fail to use evidence that is present, or cite relevant text without converting it into the final answer. This paper introduces a four-condition diagnostic protocol for evidence-utilization evaluation under matched examples, models, prompts, and scoring rules. The protocol compares no-evidence, full-context, retrieved-evidence, and oracle-evidence reference conditions, and uses Oracle-Reference Normalized Context Utilization (ONCU) as a denominator-valid estimate of recovered oracle-reference evidence advantage. The empirical study evaluates five local open-weight models from the Qwen, Gemma, Llama, and Mistral families over Controlled-ONCU-safe16K, HotpotQA-ONCU, and 2WikiMultiHopQA-ONCU, comprising 18,000 ONCU-compatible predictions. Results show a task-dependent diagnostic pattern: controlled synthetic settings expose reduced recovery when the same evidence is embedded in long input rather than supplied compactly, while realistic multi-hop reconstructions show that full-context inputs outperform the tested retrieved inputs in denominator-free answer and evidence metrics, with ONCU supporting the same direction on oracle-improving groups. Sensitivity audits with stronger retrieval settings narrow some gaps but do not overturn the scoped interpretation. The main contribution is therefore not a single utilization ratio, but a matched diagnostic protocol that separates no-evidence answerability, oracle-evidence recoverability, full-context recovery, retrieval-conditioned recovery, denominator validity, and companion answer/evidence diagnostics.

Haizhou Xia

University of Western Ontario, Canada

Question-Answering Benchmarks Answer Diagnosis

Jun 4, 2026cs.CL

Evidence Graph Consistency in Retrieval-Augmented Generation: A Model-Dependent Analysis of Hallucination Detection

Retrieval-Augmented Generation (RAG) reduces but does not eliminate hallucination in large language models. Existing detection methods rely on flat similarity between generated answers and retrieved passages, ignoring structural relationships among evidence pieces and answer claims. We propose Evidence Graph Consistency (EGC), a framework that constructs a local evidence graph per response and computes five structural consistency measures as hallucination indicators. Evaluated on the full question answering split of RAGTruth across six LLMs (5,767 responses), EGC reveals a consistent model-family split: graph consistency features show the expected diagnostic direction for hallucinations in Llama-2 models but exhibit systematic reversal in GPT-4, GPT-3.5, and Mistral-7B. This reversal suggests qualitatively different hallucination patterns across model families and indicates that embedding-based graph consistency cannot serve as a model-independent hallucination detection signal.

Jianru Shen

University of Montana Missoula, MT, USA

Citation Hallucination Detection Hallucination Detection Answer Reverse

Jun 4, 2026cs.AI

Answer Presence Drives RAG Rewriting Gains

Retrieval-augmented QA pipelines often route retrieved passages through an LLM \emph{rewriter} before a smaller reader, lifting F1 by tens of points on multi-hop benchmarks; this gain is typically credited to improved evidence quality. We ask whether that lift is causally driven by the gold answer string appearing in the rewritten context rather than by curation per se, using a controlled intervention audit. For each rewritten context we re-run the reader after one of four controlled edits to the compile output: removing the gold answer span, replacing a length-matched random non-answer span (placebo), or injecting the gold into rewrites where it was absent (at the prefix or at a midpoint sentence boundary). Across twelve completed (cell, baseline) intervention runs spanning three reader families (Qwen2.5-7B, Qwen3.5-35B, GLM-4.7), two datasets (HotpotQA, 2WikiMultihopQA), and three compiler arrangements (MA-only, MB-only, MA

+

verify), removing the gold answer drops reader F1 by

28

to

64

points beyond the length-matched placebo on paired \texttt{answer-in-compile} strata, and prepending the gold into rewrites that lacked it raises F1 by

+0.7

to

+9.7

points in

10

of

12

(cell, baseline) combinations. A companion five-sentinel audit shows the conventional single-\texttt{[MASK]} probe is itself sentinel-fragile: on 2Wiki it reports a

+4.12

~F1 ``non-leakage residual'' that flips to

-3.33

to

-7.81

~F1 under four alternative sentinels and fails an equivalence test for three of those four (

1/4

~pass). We do not propose a new rewriter or mitigation; we release the intervention runner and the sentinel panel so that other rewriter-gain claims can be tested against the same standard.

Yuejie Li, Yueying Hua, Ke Yang +8

Ant Group

Rewrite Answer Random

Jun 3, 2026cs.CL

Boosting Self-Consistency with Ranking

Self-consistency improves large language models by sampling multiple reasoning paths and selecting the most frequent answer, but majority voting often fails to recover correct answers that are already present among the samples. We address this limitation with Ranking-Improved Self-Consistency (RISC), which reformulates answer selection in self-consistency as a ranking problem. Instead of relying on a single uncertainty or confidence signal, RISC uses a lightweight LambdaRank model to score candidate answers with five carefully designed features that capture answer frequency, semantic centrality, and reasoning-trace consistency. We evaluate RISC on three datasets under a range of test-time budgets. Across datasets, RISC consistently achieves a better accuracy-efficiency trade-off than standard self-consistency and strong baselines, with particularly large gains on question answering benchmarks. Further analysis shows that the proposed features are individually useful and, more importantly, complementary, highlighting the value of learning to combine multiple informative signals for test-time answer selection.

Maria Marina, Daniil Moskovskiy, Sergey Pletenev +3

AIRI · Skoltech · EPFL

Self-Consistency Question-Answering Benchmarks Answer Reasoning Paths Multiple Centrality Measures