arXiv subjects

Jay-Yoon Lee

Publications and source records attributed to Jay-Yoon Lee.

At least 19 recordsLinked to original sources

arXiv2026

Where Search Agents Go Off the Rails: Graph-Grounded Step-Level Verification

Search-augmented reasoning systems can reach correct answers while making errors in intermediate steps. Reliable auditing must check whether evidence $D$ covers the question $Q$, reasoning $T$ addresses $Q$, and $T$ is grounded in $D$. We introduce GraphRail, which represents requirements, evidence, and reasoning as structured claims and compares them without a generative judge or gold answer. Across three multi-hop QA benchmarks and three reasoning frameworks, GraphRail improves error recall over generative verifiers while remaining stable under stochastic extraction. Its diagnoses can also guide online reasoning, providing a traceable way to detect and correct intermediate errors.

cs.AI↗

arXiv2026

Subtraction Gets You More: Gap-Aware Retrieval for Multimodal Multi-Hop QA

In multimodal multi-hop question answering, we focus on the initial retrieval stage via two distinct tasks: (1) evidence set completion, retrieving missing evidence given context, and (2) sequential pool construction, iteratively building the top-$K$ pool from the scratch. Under these settings, we point out that conventional iterative retrieval frameworks often suffer from Semantic Anchoring, where previously fetched evidence traps the retriever and yields entity-centric redundancy. To break this trap, we propose GRAIL (Gap-aware Retrieval via Adaptive Implicit Localization), a paradigm that performs implicit query rewriting directly at the embedding level. By context-subtractive query steering, GRAIL excels at compositional cross-modal reasoning, while additive embedding updates show strength on localized information aggregation. To leverage these complementary strengths, we propose a Task-Adaptive Hybrid Framework, achieving a 35.7% macro-averaged performance gain on the MultimodalQA evidence set completion task. Extensive evaluations demonstrate that sequential GRAIL retrieves in a superior, noise-resilient manner, significantly expanding the search horizon through iterative gap-aware optimization.

cs.IR↗

arXiv2026

Cross-Agent Learning Signals Enable Coordinated Role-Decomposed LLM Training

Agentic search systems must coordinate evidence acquisition and response generation, yet existing approaches either couple both roles under a single agent objective or decompose them without disentangling their respective contributions to the final outcome. We introduce DAC (Divide and Cooperate), a role-decomposed training framework that, given task-specific external verification signals, trains a searcher and a generator with role-specific cross-verification rewards. DAC allows the generator to abstain when the retrieved evidence appears insufficient, and uses this decision together with externally evaluated search sufficiency to assign appropriate credit to each role. To prevent degenerate over-abstention, we further introduce hard-positive evidence augmentation, which discourages abstaining on sufficient but hard evidence. Across seven general and multi-hop QA benchmarks and two model backbones, DAC consistently outperforms strong single-agent and multi-agent baselines. Controlled evaluations show that these gains arise from coordinated improvements in both search and generation. Our results highlight the importance of explicitly and properly assigning credit across interacting roles when training agentic search systems.

cs.LG↗

arXiv2026

LaSEr-Edit: Localized Span-level Error Editing with Energy-based Localization

As large language models (LLMs) are widely adopted in real-world applications, it has become critical to ensure LLMs satisfy safety constraints, such as non-toxicity and logical consistency, as well as task- and situation-specific constraints. Controlling the output through instructions is a simple and tempting approach; however, it remains brittle, is opaque in how it influences model behavior, and thus cannot reliably ensure constraint satisfaction. Moreover, most recent controlled text generation (CTG) methods require access to the internal components of language models--such as weights or logits--making them incompatible with popular API-based LLMs. In this work, we propose LaSEr-Edit, a constraint-satisfying text revision method that can be applied to any LLMs, black- or white-box. We first find that lightweight, task-specific energy-based models (EBMs) achieve error-localization performance competitive with or even better than that of much larger LLMs, while operating substantially faster. Based on this finding, we propose two variants of text revision methods that incorporate energy-based error localization: LaSEr-LLM Edit, which instructs an LLM to edit text given EBM-predicted error spans, and LaSEr-EBM Edit, which uses the EBM not only for localization but also for editing by reranking edit candidates. Through experiments in diverse single-constraint control tasks, we show that LaSEr-LLM Edit controls text better than plain LLM-based editing in most of the tasks. We also find that LaSEr-EBM Edit further improves the control performance of LaSEr-LLM Edit and achieves among the strongest controllability across all tasks. Furthermore, we find that LaSEr-Edit, especially LaSEr-EBM Edit, performs well even when multiple constraints are controlled simultaneously.

cs.CL↗

arXiv2026

Scoring a Set, Not Summing Passage Scores: Effective and Efficient Set Retrieval

Multi-hop question answering requires retrieving multiple evidence passages whose usefulness often depends on one another. Conventional retrievers either rank passages independently or construct evidence sequentially through locally supervised next-passage decisions. Sequential conditioning captures some cross-passage dependencies, but its local extension scores do not provide a common criterion for comparing complete evidence sets of different compositions and sizes. Existing multi-hop retrievers therefore avoid directly learning a query--set compatibility function over complete evidence sets, as the exponentially large set space is daunting to cover during learning and impractical to search at inference time. To overcome these challenges, we formulate multi-hop retrieval by directly ranking candidate evidence sets with an energy-based query--set compatibility score $s_θ(q,S)$, learned from informative contrasts without exhaustive coverage or normalization of the combinatorial set space. Given a gold evidence set, we automatically generate contrasts by adding, removing, or replacing passages, yielding rich set-level supervision without additional annotation. To make search over this combinatorial space practical at inference time, we introduce a novel retrieve-and-rerank framework over the set space: ParaSet, a lightweight scorer over precomputed passage representations for efficient set exploration, and SetCE, an expressive cross-encoder for set reranking. Our experiments show that set-level retrieval consistently provides a complementary signal to passage-level relevance, becoming relatively more effective when more hops are required to reach evidence from the query. Motivated by this complementarity, combining the two signals further improves downstream QA performance, outperforming both deeper passage-level retrieval and an ensemble of distinct passage-level retrievers.

cs.IR↗

arXiv2026

OBJECTION! Lawyer Agents Mitigate Guilty Bias in Legal Judgment Prediction

Legal Judgment Prediction (LJP) models are typically trained on documents that describe facts from a prosecutorial perspective. Existing datasets further exhibit severe label imbalance toward guilty outcomes. Consequently, these models suffer from "Guilty Bias", blindly accepting the prosecution's narrative as objective truth. Previous studies employing three-step reasoning structures or training on synthetically generated innocence data improve overall accuracy, but they still fail to mitigate bias at inference time. In this paper, we introduce OBJECTION, an inference-time pipeline that integrates an Adversarial Lawyer Agent into each 3-step reasoning of offense, unlawfulness, and culpability. Unlike generic critics, our agent actively challenges the model's presumptions of guilt by injecting legal defense arguments at each reasoning stage. To thoroughly evaluate this, we present a new "Natural Innocent" dataset including 3.4k real-world cases, overcoming the limitations of synthetic innocence benchmarks. Test results show that OBJECTION drastically reduces the False Guilty Rate (FGR) from 82.93% (SOTA baseline) to 16.69%, proving its capability to perform substantive legal reasoning. This work denotes a key progress toward aligning Legal AI with the presumption of innocence.

cs.CL↗

arXiv2026

RARE: Redundancy-Aware Retrieval Evaluation Framework for High-Similarity Corpora

Existing QA benchmarks typically assume distinct documents with minimal overlap, yet real-world retrieval-augmented generation (RAG) systems operate on corpora such as financial reports, legal codes, and patents, where information is highly redundant and documents exhibit strong inter-document similarity. This mismatch undermines evaluation validity: retrievers can be unfairly undervalued even when they retrieve documents that provide sufficient evidence, because redundancy across documents is not accounted for in evaluation. On the other hand, retrievers that perform well on standard benchmarks often generalize poorly to real-world corpora with highly similar and redundant documents. We present RARE (Redundancy-Aware Retrieval Evaluation), a framework for constructing realistic benchmarks by (i) decomposing documents into atomic facts to enable precise redundancy tracking and (ii) enhancing LLM-based data generation with CRRF. RAG benchmark data usually requires multiple quality criteria, but LLMs often yield trivial outputs. CRRF scores criteria separately and fuses decisions by rank, improving the reliability of generated data. Applying RARE to Finance, Legal, and Patent corpora, we introduce RedQA, where a strong retriever baseline drops from 66.4% PerfRecall@10 on 4-hop General-Wiki to 5.0-27.9% PerfRecall@10 at 4-hop depth, revealing robustness gaps that current benchmarks fail to capture. RARE enables practitioners to build domain-specific RAG evaluations that faithfully reflect real-world deployment conditions.

cs.CL↗

arXiv2026

Generalizing Numerical Reasoning in Table Data through Operation Sketches and Self-Supervised Learning

Numerical reasoning over expert-domain tables often exhibits high in-domain accuracy but limited robustness to domain shift. Models trained with supervised fine-tuning (SFT) on specific datasets tend to rely on header-operation shortcuts rather than structural reasoning. We introduce TaNOS, a continual pre-training framework comprising three components: (i) header anonymization to reduce lexical memorization, (ii) operation sketches that provide minimal structural cues, and (iii) self-supervised pretraining that constructs correctness-guaranteed program-question pairs from given tables in a program-first manner. By decoupling domain semantics and numerical operation structure, TaNOS improves the transferability of numerical reasoning. Applied to an 8B instruction-tuned model, TaNOS achieves 80.13% execution accuracy on FinQA with only 10% train data, outperforming SFT baseline (73.97%) with full train data and proprietary models such as GPT-5, Gemini-2.5-Pro. Furthermore, in the domain-shift experiments, TaNOS displays nearly-negligible cross-domain gap (<2pp) when standard SFT shows over 10pp gap. These results suggest that structural guidance with operation sketches, header-agnostic representations, and correctness-guaranteed self-supervision can improve the robustness of numerical reasoning across diverse expert-domain tables.

cs.LG↗

arXiv2026

ToolMATH: A Diagnostic Benchmark for Long-Horizon Tool Use under Systematic Tool-Catalog Constraints

We introduce \ToolMATH, a math-grounded diagnostic benchmark for evaluating long-horizon tool use under controllable tool-catalog conditions. \ToolMATH converts stepwise MATH solutions into reusable Python tools with natural-language descriptions and typed schemas, and pairs each problem with a tool environment requiring sequential tool use, intermediate-output reuse, and logically connected tool-call chains. \ToolMATH controls tool availability and catalog difficulty by constructing gold tools and graded distractors with varying similarity to gold tools. \ToolMATH also incorporates behavior-conditioned metrics, enabling diagnostic evaluation beyond final accuracy. Building on these measurements, \ToolMATH emphasizes three evaluation axes: (1) \emph{Adaptability} measures how much Gold-only success is retained when gold tools are replaced entirely by distractors; (2) \emph{Robustness} measures stability under adding distractors as a noise; and (3) \emph{Tool Connectivity} measures whether models preserve accuracy over long executed tool-call chains. Furthermore, trace-level failure analyses characterize how models fail under each tool-catalog condition. Together, these diagnostics reveal distinct model profiles: reliable tool use, tool avoidance, adaptive substitution, and impacts of unreliable tool catalogs. Overall, \ToolMATH provides a controlled testbed for evaluating how language models adapt to changing tool availability, remain robust to distractors, and maintain correctness across long-horizon tool-use trajectories.

cs.CL↗

arXiv2026

Shallow Prefill, Deep Decoding: Efficient Long-Context Inference via Layer-Asymmetric KV Visibility

Long-context inference in decoder-only language models is costly because long prompts are processed during Prefill, cached at every layer, and repeatedly attended to during autoregressive Decode. We introduce \emph{Shallow Prefill, dEEp Decode} (SPEED), a phase-asymmetric KV-visibility policy that materializes non-anchor prompt-token KV states only in lower layers while keeping Decode-phase tokens full-depth. Unlike previous approaches that make upper-layer prompt KV states cheaper to store or construct, SPEED removes prefill tokens from the upper-layer Decode visibility set altogether. With a minimal BoS anchor, this simple change preserves broad benchmark quality while reducing long-context cost. In a controlled Llama-3.1-8B instruction-tuning study, SPEED using only 75\% of layers for prefill tokens reaches 51.2 average score on OLMES-style benchmarks, compared with 51.4 for the full-depth baseline, while improving TTFT by 33\%, TPOT by 22\%, and reducing active KV memory by 25.0\% at 128K context. Layer-wise diagnostics suggest that this cutoff retains the main prompt-selection and representation-stabilization regions of the full-depth model. These results show that long-context prompt tokens need not always persist as full-depth KV-cache objects when Decode-phase tokens remain full-depth.

cs.AI↗

arXiv2026

Latent Self-Consistency for Reliable Majority-Set Selection in Short- and Long-Answer Reasoning

Probabilistic decoding in Large Language Models (LLMs) often yields inconsistent outputs, particularly on complex or long-form questions. Self-Consistency (SC) mitigates this for short-form QA by majority voting over exact strings, whereas Universal Self-Consistency (USC) and Weighted Unigram Consistency Score (WUCS) extend to long-form responses but lose accuracy on short-form benchmarks. We introduce \textbf{Latent Self-Consistency (LSC)}, which selects the most semantically consistent response using learnable token embeddings. LSC's lightweight forward processing of summary tokens only introduces negligible runtime overhead (at most $0.9\%$) on top of standard decoding of the base LLM, and requires no changes to the model architecture. Across 6 short-form and 5 long-form reasoning benchmarks (e.g., MATH, MMLU, TruthfulQA), LSC surpasses SC, USC, and WUCS on both short-form and long-form on average performance, while adding negligible computational overhead on vanilla inference. These results position LSC as a reliable consistency-selection method that works effectively across various answer formats. Additionally, LSC provides well-calibrated confidence estimates, maintaining low expected calibration error across both answer formats.

cs.CL↗

arXiv2026

SEAL-pose: Enhancing 3D Human Pose Estimation via a Learned Loss for Structural Consistency

3D human pose estimation (HPE) is characterized by intricate local and global dependencies among joints. Conventional supervised losses are limited in capturing these correlations because they treat each joint independently. Previous studies have attempted to promote structural consistency through manually designed priors or rule-based constraints; however, these approaches typically require manual specification and are often non-differentiable, limiting their use as end-to-end training objectives. We propose SEAL-pose, a data-driven framework in which a learnable loss-net trains a pose-net by evaluating structural plausibility. Rather than relying on hand-crafted priors, our joint-graph-based design enables the loss-net to learn complex structural dependencies directly from data. Extensive experiments on three 3D HPE benchmarks with eight backbones show that SEAL-pose reduces per-joint errors and improves pose plausibility compared with the corresponding backbones across all settings. Beyond improving each backbone, SEAL-pose also outperforms models with explicit structural constraints, despite not enforcing any such constraints. Finally, we analyze the relationship between the loss-net and structural consistency, and evaluate SEAL-pose in cross-dataset and in-the-wild settings.

cs.CV↗

arXiv2025

GraphCheck: Multipath Fact-Checking with Entity-Relationship Graphs

Automated fact-checking aims to assess the truthfulness of textual claims based on relevant evidence. However, verifying complex claims that require multi-hop reasoning remains a significant challenge. We propose GraphCheck, a novel framework that transforms claims into entity-relationship graphs for structured and systematic fact-checking. By explicitly modeling both explicit and latent entities and exploring multiple reasoning paths, GraphCheck enhances verification robustness. While GraphCheck excels in complex scenarios, it may be unnecessarily elaborate for simpler claims. To address this, we introduce DP-GraphCheck, a variant that employs a lightweight strategy selector to choose between direct prompting and GraphCheck adaptively. This selective mechanism improves both accuracy and efficiency by applying the appropriate level of reasoning to each claim. Experiments on the HOVER and EX-FEVER datasets demonstrate that our approach outperforms existing methods in verification accuracy, while achieving strong computational efficiency despite its multipath exploration. Moreover, the strategy selection mechanism in DP-GraphCheck generalizes well to other fact-checking pipelines, highlighting the broad applicability of our framework.

cs.CL↗

arXiv2025

Stop-RAG: Value-Based Retrieval Control for Iterative RAG

Iterative retrieval-augmented generation (RAG) enables large language models to answer complex multi-hop questions, but each additional loop increases latency, costs, and the risk of introducing distracting evidence, motivating the need for an efficient stopping strategy. Existing methods either use a predetermined number of iterations or rely on confidence proxies that poorly reflect whether more retrieval will actually help. We cast iterative RAG as a finite-horizon Markov decision process and introduce Stop-RAG, a value-based controller that adaptively decides when to stop retrieving. Trained with full-width forward-view Q($λ$) targets from complete trajectories, Stop-RAG learns effective stopping policies while remaining compatible with black-box APIs and existing pipelines. On multi-hop question-answering benchmarks, Stop-RAG consistently outperforms both fixed-iteration baselines and prompting-based stopping with LLMs. These results highlight adaptive stopping as a key missing component in current agentic systems, and demonstrate that value-based control can improve the accuracy of RAG systems.

cs.LG↗

arXiv2025

BridG MT: Enhancing LLMs' Machine Translation Capabilities with Sentence Bridging and Gradual MT

Recent Large Language Models (LLMs) have demonstrated impressive translation performance without requiring fine-tuning on additional parallel corpora. However, they still face significant challenges in certain scenarios, particularly when translating low-resource languages. A common approach to address this issue is to provide external knowledge, such as few-shot examples, to assist LLMs in translating specific source sentences. However, this method is fundamentally limited by the quality or quantity of relevant sources, which cannot always be guaranteed. To reduce LLMs' reliance on external sources, we propose BridG MT, a method that combines Sentence Bridging, which generates a sequence of sentences as a bridge that gradually transition from easy-to-translate to more difficult, and Gradual MT, which sequentially translates these sentences using earlier translations as few-shot examples for subsequent ones. Experiments conducted on four LLMs across seven languages demonstrate that our method effectively enhances translation performance, even outperforming translation methods that rely on a large number of few-shot examples.

cs.CL↗

arXiv2025

Introducing Verification Task of Set Consistency with Set-Consistency Energy Networks

Examining logical inconsistencies among multiple statements (such as collections of sentences or question-answer pairs) is a crucial challenge in machine learning, particularly for ensuring the safety and reliability of models. Traditional methods that rely on pairwise comparisons often fail to capture inconsistencies that only emerge when more than two statements are evaluated collectively. To address this gap, we introduce the task of set-consistency verification, an extension of natural language inference (NLI) that assesses the logical coherence of entire sets rather than isolated pairs. Building on this task, we present the Set-Consistency Energy Network (SC-Energy), a novel model that employs a contrastive loss framework to learn the compatibility among a collection of statements. Our approach not only efficiently verifies inconsistencies and pinpoints the specific statements responsible for logical contradictions, but also significantly outperforms existing methods including prompting-based LLM models. Furthermore, we release two new datasets: Set-LConVQA and Set-SNLI for set-consistency verification task.

cs.LG↗

arXiv2025

Mind the Gap: Aligning the Brain with Language Models Requires a Nonlinear and Multimodal Approach

Self-supervised language and audio models effectively predict brain responses to speech. However, traditional prediction models rely on linear mappings from unimodal features, despite the complex integration of auditory signals with linguistic and semantic information across widespread brain networks during speech comprehension. Here, we introduce a nonlinear, multimodal prediction model that combines audio and linguistic features from pre-trained models (e.g., LLAMA, Whisper). Our approach achieves a 17.2% and 17.9% improvement in prediction performance (unnormalized and normalized correlation) over traditional unimodal linear models, as well as a 7.7% and 14.4% improvement, respectively, over prior state-of-the-art models. These improvements represent a major step towards future robust in-silico testing and improved decoding performance. They also reveal how auditory and semantic information are fused in motor, somatosensory, and higher-level semantic regions, aligning with existing neurolinguistic theories. Overall, our work highlights the often neglected potential of nonlinear and multimodal approaches to brain modeling, paving the way for future studies to embrace these strategies in naturalistic neurolinguistics research.

cs.CL↗

arXiv2024

Comparing Neighbors Together Makes it Easy: Jointly Comparing Multiple Candidates for Efficient and Effective Retrieval

A common retrieve-and-rerank paradigm involves retrieving relevant candidates from a broad set using a fast bi-encoder (BE), followed by applying expensive but accurate cross-encoders (CE) to a limited candidate set. However, relying on this small subset is often susceptible to error propagation from the bi-encoders, which limits the overall performance. To address these issues, we propose the Comparing Multiple Candidates (CMC) framework. CMC compares a query and multiple embeddings of similar candidates (i.e., neighbors) through shallow self-attention layers, delivering rich representations contextualized to each other. Furthermore, CMC is scalable enough to handle multiple comparisons simultaneously. For example, comparing ~10K candidates with CMC takes a similar amount of time as comparing 16 candidates with CE. Experimental results on the ZeSHEL dataset demonstrate that CMC, when plugged in between bi-encoders and cross-encoders as a seamless intermediate reranker (BE-CMC-CE), can effectively improve recall@k (+4.8%-p, +3.5%-p for R@16, R@64) compared to using only bi-encoders (BE-CE), with negligible slowdown (<7%). Additionally, to verify CMC's effectiveness as the final-stage reranker in improving top-1 accuracy, we conduct experiments on downstream tasks such as entity, passage, and dialogue ranking. The results indicate that CMC is not only faster (11x) but also often more effective than CE, with improved prediction accuracy in Wikipedia entity linking (+0.7%-p) and DSTC7 dialogue ranking (+3.3%-p).

cs.CL↗