Large language models (LLMs) can generate fluent reasoning traces that nevertheless lead to incorrect answers, making response-level uncertainty estimation important for abstention, human review, and adaptive compute allocation. Existing approaches generally fall into three categories: passive single-trace methods use token-level confidence signals, sampling-based methods compare multiple complete traces at higher generation cost, and active prefix-based methods probe partial traces to study answer stabilization or preference transitions. However, none actively re-elicits an answer from a completed reasoning trace to measure its consistency with and support for the original answer. To address this gap, we introduce Trace-Conditioned Answer Consistency (TrAC), a correctness-supervised uncertainty quantification framework that combines active and passive signals anchored to one completed reasoning trace. Its active component, Prefix-Conditioned Elicitation (PCE), re-elicits a short answer conditioned on the completed trace and represents both its consistency with the original answer and its token-level probabilistic support. Its passive component, Trace Uncertainty Profile (TUP), summarizes how token-level uncertainty evolves throughout the original generation without additional decoding. A lightweight head then integrates the two representations into a response-correctness score. Across five mathematical reasoning benchmarks and three LLM families, TrAC improves macro AUROC by 1.8% and reduces AURC by 3.4% relative to eight-sample self-consistency, while using one complete reasoning trace and a short cached answer probe. When eight samples are already available, augmenting sample consensus with re-elicitation further improves macro AUROC by 4.3% and reduces AURC by 8.3%, without additional full-trace generation.
Although large language models (LLMs) can produce fluent spatial reasoning traces, their intermediate relations may fail to support the final conclusion, making token-level confidence insufficient for final-answer reliability estimation. Existing formal verifiers provide stronger semantic evidence, but their applicability is partial: a parsed claim need not yield a definite semantic verdict. To address this issue, we introduce SymboUQ, a symbolic uncertainty quantification framework that estimates final-answer reliability from reasoning traces by distinguishing symbolizability, whether a claim can be represented in the verifier's formal language, from semantic determinacy, whether its execution yields an entailed or contradicted verdict rather than an unknown or not-evaluable outcome. SymboUQ comprises (i) a Layout Auditor that executes ordered spatial claims and extracts feasibility, conflict, and repair evidence; (ii) a label-free Determinacy Profile that characterizes effective executable coverage; and (iii) a Determinacy-Aware Reliability Composer that integrates constraint-based, representation-based, and decoding-based scores according to verifier applicability. Extensive experiments on five spatial reasoning benchmarks with four frozen LLM backbones show that SymboUQ achieves approximately an 8% relative improvement in AUROC and a 7% relative reduction in class-balanced Brier loss over the strongest baseline.
Dahai Yu, Lin Jiang, Rongchao Xu et al.· 0 citations
Large language models (LLMs) remain vulnerable to answer anchoring, overconfidence, and fluent but unsupported justifications in binary and multiple-choice question answering. Recent answer-first debate frameworks force agents to inspect alternative stances, but they still spend substantial budget defending likely wrong answers and let a final judge consume long free-form interaction traces. We propose TRACE-QA, a training-free multi-agent protocol that routes each instance to a sparse set of reasoning operators, constructs option-blind necessity constraints, audits every candidate in a structured elimination ledger, revisits risky eliminations through global risk-aware rescue, and aggregates role-specialized judge votes. Under a unified three-seed protocol on HoVer, BoolQ, CosmosQA, and CommonsenseQA, TRACE-QA attains an average of 0.7731, exceeding the strongest common-protocol baseline (CFMAD) by 1.94 percentage points and improving on all five evaluation columns. Risk-aware rescue lowers the false-elimination rate from 5.2% without rescue to 1.4% and recovers 73.1% of recoverable false eliminations. We further evaluate structured graph reasoning on GraphSQA and a discrete in-toolset subset of Graph4real: TRACE-QA improves over text-only baselines, and with deterministic graph tools reaches 95.1%–97.8% accuracy across Graph4real scales. Full TRACE-QA remains substantially more expensive than chain-of-thought, but uses 19.7% fewer total tokens than CFMAD, and an adaptive variant saves 29.5% of its tokens at a 0.67-point accuracy cost. These results support constraint-guided elimination as an auditable alternative to answer-first debate, while exposing residual dependence on routing quality, ledger reliability, and backbone capability.
Jiaxin Lu, Hao Chen, Yangyan Zhu et al.· Journal of King Saud Univers...· 0 citations
Large language models (LLMs) have increasingly supported response generation grounded in user-provided knowledge spanning heterogeneous structures. However, existing benchmarks provide limited assessment of whether LLMs can faithfully perform multi-hop reasoning chains across such knowledge contexts while remaining robust to variations in their input order. We introduce TKFQA, a factuality consistency benchmark comprising 10,130 question-answering (QA) pairs grounded in tables, texts, and knowledge graphs (KGs). Each example is constructed from an explicit counterfactual reasoning chain, enabling the joint evaluation of answer correctness, reasoning-chain accuracy, and robustness to different input-order. An extensive evaluation of 14 open- and closed-source LLMs reveals that state-of-the-art models exhibit limited reasoning-chain accuracy and remain sensitive to variations in the input order of heterogeneous knowledge contexts. To address these limitations, we propose ORLF, an LLM-agnostic training framework that models cross-context topological relations through knowledge-specific latent vectors. ORLF integrates context-wise position encoding, a latent-bridge attention mask, and topological knowledge bias to preserve knowledge-specific bias and encode topological semantics. Experiments across four LLM backbones show that ORLF outperforms competitive training-free and LoRA-based baselines, improving average Exact Match and Reasoning-Chain Accuracy by 2.15% and 4.29%, respectively, while reducing order-induced performance standard deviation by 0.04% to 3.01%.
Shibo Chu, Yuze Liu, Tiehua Zhang et al.· 0 citations
Tool-using large language model (LLM) agents read untrusted content, maintain memory, delegate tasks, and invoke tools with external side effects. Terminal attack-success or policy-violation rates do not show what happens between exposure and commit or whether a defense also suppresses authorized actions. We introduce ContainmentBench, a sandboxed benchmark comprising a 504-scenario specification dataset, a shared rollout-trace schema, and stage-scoped metrics for endpoint violations, logged propagation, and explicitly authorized taint-exposed proposals that commit. The main Qwen2.5-7B-Instruct study evaluates seven policy conditions and five seeds, yielding a 17,640-record trace corpus. Across 600 matched active-tainted rollout pairs, no committed policy violation was observed under either taint-only or intent-ledger enforcement. Their execution records nevertheless differed: 441 pairs (73.5%) had different values in a shared 12-field trace summary that includes commit-related diagnostics, and the mean authorized proposal-commit score was 0.164 under taint-only enforcement and 0.857 under intent-ledger enforcement, compared with 0.923 under tool-boundary enforcement. Logged-propagation rankings changed with stage selection and normalization. In a limited set of custom AgentDojo-native workflows, committed violations were observed without defense and were not observed under either evaluated defense. A separate 6,048-rollout Mistral/common-JSON model-interface configuration retained the v1-to-v2 proposal-commit improvement, but committed violations were observed under intent-ledger v2. Equal terminal outcomes do not imply equal containment. The evaluation uses synthetic workflows. The intended intent-ledger mechanism assumes schema-aligned authorization metadata; one public-status task family violates this assumption and is analyzed separately.
With the rapid advancement of large language models (LLMs), modern systems not only possess strong foundational capabilities and extensive knowledge, but can also solve complex problems via long, multi-step reasoning. However, as reasoning traces become longer, LLMs may produce a substantial amount of hallucinated content during the reasoning process, which is often difficult to detect. In this work, we conduct a fine-grained analysis of hallucinations arising in LLM reasoning and find that the reasoning traces are particularly prone to Context-Sensitive Factual Hallucinations: cases where the model actually has the relevant knowledge, yet makes factual errors due to contextual interference during reasoning. To address this issue, we propose Step-level Self-Consistency Group Relative Policy Optimization (SSC-GRPO), which assigns step-level rewards to reasoning traces by computing self-consistency scores of individual steps across multiple rollouts. Compared with prior methods, SSC-GRPO achieves state-of-the-art performance on both mathematical reasoning benchmarks and hallucination leaderboards. Our results offer a new perspective for detecting and mitigating hallucinations in the reasoning process of large language models.
Xiaomeng Hu, Jiaqi Hu, Hao Chen et al.· 0 citations
Text-to-SQL systems are commonly evaluated using ground-truth SQL queries or reference execution results, but such supervision is unavailable at inference time in real-world deployments. This creates a critical verification problem: given only a user question, database context, and generated SQL, can a system estimate whether the generated query is likely to correctly answer the question? Recent approaches use LLMs as judge or specialized agents to inspect generated SQL, but their decisions can be difficult to trace. Outcome Reward Models (ORMs) address this by learning from execution-labeled candidate SQLs and assigning correctness scores to unseen queries, yet they still provide limited visibility into the signals behind each verification. To address this limitation, we propose TraceSQL, a lightweight and traceable verification model built on explicit diagnostic features. TraceSQL combines 67 features capturing question ambiguity, question requirements, question-schema-SQL consistency, SQL structure, and intent alignment. These signals remain available for examining which factors influence each prediction and for tracing decisions back to diagnostic evidence. On BIRD development databases, TraceSQL achieves 66.47% F1 and 64.48% ROC-AUC, compared with 61.87% F1 and 58.26% ROC-AUC for the GradeSQL-7B ORM baseline on the same generated-SQL evaluation. Feature attribution further shows that the model relies on both semantic grounding and deterministic SQL-structure signals. These results show that SQL verification can be performed with a lightweight learned model while retaining feature-level evidence for inspecting and diagnosing its predictions.
N. Shukla, Debasmita Panda, Srutanik Bhaduri et al.· 0 citations