Searcharxiv⌕ Search

arXiv subjects

Yuelyu Ji

Publications and source records attributed to Yuelyu Ji.

At least 19 recordsLinked to original sources
arXiv2026

SPEAR: Distilling Domain-Adaptive Reasoning Skeletons via Sequential Symbolic Alignment in Reinforcement Learning

Reinforcement learning-based knowledge distillation has the potential to transfer complex reasoning from teacher to student models, yet it currently faces a critical dilemma: researchers must choose between sparse outcome-based rewards, which provide insufficient logical guidance, or expensive neural Process Reward Models (PRMs) for dense signals. We resolve this by introducing SPEAR (Symbolic Process Evaluation and Alignment Reward), a training-free and plug-and-play process reward method for sequence-level on-policy distillation. SPEAR projects natural-language reasoning traces into domain-adaptive symbolic milestones, providing an efficient proxy for process-level reasoning alignment. By utilizing the longest common subsequence (LCS) to align student explorations with teacher milestones, SPEAR provides a dense, order-aware reward signal that enforces logical consistency without the need for an external neural verifier. Our experiments across math, science, and commonsense reasoning tasks demonstrate that SPEAR effectively bridges the reasoning gap between student and teacher models via sequence-level distillation with efficient dense process rewards. Our code and data are available at: https://github.com/zhuochunli/SPEAR.

cs.CL↗
arXiv2026

MedRAGChecker: Claim-Level Verification for Biomedical Retrieval-Augmented Generation

Biomedical retrieval-augmented generation (RAG) can ground LLM answers in medical literature, yet long-form outputs often contain isolated unsupported or contradictory claims with safety implications. We introduce MedRAGChecker, a claim-level verification and diagnostic framework for biomedical RAG. Given a question, retrieved evidence, and a generated answer, MedRAGChecker decomposes the answer into atomic claims and estimates claim support by combining evidence-grounded natural language inference (NLI) with biomedical knowledge-graph (KG) consistency signals. Aggregating claim decisions yields answer-level diagnostics that help disentangle retrieval and generation failures, including faithfulness, under-evidence, contradiction, and safety-critical error rates. To enable scalable evaluation, we distill the pipeline into compact biomedical models and use an ensemble verifier with class-specific reliability weighting. Experiments on four biomedical QA benchmarks show that MedRAGChecker reliably flags unsupported and contradicted claims and reveals distinct risk profiles across generators, particularly on safety-critical biomedical relations.

cs.CL↗
arXiv2026

More Debate, Same Evidence: Structural Limits of Homogeneous Multi-Agent Groundedness

Large language model (LLM) judges are increasingly organized as multi-agent panels under the assumption that exchanging critiques improves judgment quality. We test this assumption for \emph{groundedness verification}, where a judge must determine whether a claim is supported by the supplied evidence. We evaluate a homogeneous three-agent panel on six public fact-verification and hallucination-detection benchmarks. Relative to a fixed single-agent reference, the panel's system-level accuracy difference ranges from $+8.5$ to $-4.4$ percentage points: two datasets show reliable gains, one shows a reliable loss, and three are statistically inconclusive. Because the reference and panel use different model variants, these differences characterize the complete systems rather than isolate a causal debate effect.

cs.AI↗
arXiv2026

Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry

Large language models (LLMs) are widely used as reference-free evaluators via prompting, but this "LLM-as-a-Judge" paradigm is costly, opaque, and sensitive to prompt design. In this work, we investigate whether smaller models can serve as efficient evaluators by leveraging internal representations instead of surface generation. We uncover a consistent empirical pattern: small LMs, despite with weak generative ability, encode rich evaluative signals in their hidden states. This motivates us to propose the Semantic Capacity Asymmetry Hypothesis: evaluation requires significantly less semantic capacity than generation and can be grounded in intermediate representations, suggesting that evaluation does not necessarily need to rely on large-scale generative models but can instead leverage latent features from smaller ones. Our findings motivate a paradigm shift from LLM-as-a-Judge to Representation-as-a-Judge, a decoding-free evaluation strategy that probes internal model structure rather than relying on prompted output. We instantiate this paradigm through INSPECTOR, a probing-based framework that predicts aspect-level evaluation scores from small model representations. Experiments on reasoning benchmarks (GSM8K, MATH, GPQA) show that INSPECTOR substantially outperforms prompting-based small LMs and closely approximates full LLM judges, while offering a more efficient, reliable, and interpretable alternative for scalable evaluation. The code and data are available at: https://github.com/zhuochunli/Representation-as-a-judge

cs.CL↗
arXiv2026

What Makes a Medical Checker Trainable? Diagnosing Signal Collapse and Reward Hacking in Checker-Guided RAG for Biomedical QA

Medical RAG needs evidence-grounded claims, so plugging a claim-level NLI checker into retrieval-augmented RL is intuitive. \textbf{We find that the checker's \emph{output distribution} during training, not its held-out accuracy, decides whether it provides trainable gradient.} We compare four NLI checker back-ends as process rewards inside a GRPO-trained medical RAG agent (Qwen2.5-7B, replicated on Qwen3-4B and Llama-3.1-8B) across four held-out medical QA benchmarks. Three diagnostic findings emerge. \textbf{(i)} Signal collapse is log-prob-specific: LLM log-probability scoring labels over 97\% of claims neutral -- collapsing the RL gradient to zero -- while a calibrated MedNLI classifier scores the same pairs non-degenerately. \textbf{(ii)} Moderate signal beats strong signal on answer quality: a strong proprietary checker triggers a three-step reward-hacking cascade -- ultra-short answers, search avoidance, language collapse -- so a moderate-signal local classifier trains a higher-quality model (\textbf{+12\% BERTScore over zero-shot, no GPT dependency}). \textbf{(iii)} Signal strength is policy-dependent: the same checker registers as moderate on one policy but strong on another without triggering the cascade end-state. We frame these as boundary conditions for verifier-as-reward systems.

cs.CL↗
arXiv2026

StepGap: A Hybrid NLI-LLM Checker for Step-Level Evidence-Gap Detectionin Multi-Hop Question Answering

We present \textbf{StepGap}, a hybrid NLI-LLM decision tree that detects step-level evidence gaps in multi-hop QA and emits one of three typed labels: \textsc{Contradicted Claim} (CC), \textsc{Irrelevant Evidence} (IE), or \textsc{Missing Bridge} (MB), each tied to a concrete repair action. On 82 multi-hop questions (181 annotated steps, $κ{=}0.704$), StepGap reaches sF1$=$72.0, within the bootstrap confidence interval of an LLM-only baseline (70.1) but with a more decomposable structure: every StepGap stage \emph{hurts} F1 when removed, while three of four LLM-only removals \emph{improve} F1 -- a sign of \emph{competing-error cancellation}, where internal stages mask each other's errors. We further expose a \emph{Q-F1 trap}: question-level F1 is mechanically inflated by checkers that flag every step, making step-level F1 the necessary diagnostic. Used as a typed GRPO process reward, StepGap improves Qwen2.5-7B-Instruct Exact Match from $32.1{\pm}0.3$ to $35.4{\pm}0.9$ across three seeds, with the single-run comparison showing a $+5.6$ Avg EM gain over the matched Search-R1 GRPO reproduction.

cs.CL↗
arXiv2026

A Scoping Review of LLM-as-a-Judge in Healthcare and the MedJUDGE Framework

As large language models (LLMs) increasingly generate and process clinical text, scalable evaluation has become critical. LLM-as-a-Judge (LaaJ), which uses LLMs to evaluate model outputs, offers a scalable alternative to costly expert review, but its healthcare adoption raises safety and bias concerns. We conducted a PRISMA-ScR scoping review of six databases (January 2020-January 2026), screening 11,727 studies and including 49. The landscape was dominated by evaluation and benchmarking applications (n=37, 75.5%), pointwise scoring (n=42, 85.7%), and GPT-family judges (n=36, 73.5%). Despite growing adoption, validation rigor was limited: among 36 studies with human involvement, the median number of expert validators was 3, while 13 (26.5%) used none. Risk of bias testing was absent in 36 studies (73.5%), only 1 (2.0%) examined demographic fairness, and none assessed temporal stability or patient context. Deployment remained limited, with 1 study (2.0%) reaching production and four (8.2%) prototype stage. Importantly, these gaps may interact: when judges and evaluated systems share training data or architectures, they may inherit similar blind spots, and agreement metrics may fail to distinguish true validity from shared errors. Minimal human oversight, limited bias assessment, and model monoculture together represent a governance gap where current validation may miss clinically significant errors. To address this, we propose MedJUDGE (Medical Judge Utility, De-biasing, Governance and Evaluation), a risk-stratified three-pillar framework organized around validity, safety, and accountability across clinical risk tiers, providing deployment-oriented evaluation guidance for healthcare LaaJ systems.

cs.CY↗
arXiv2026

Scaling Medical Reasoning Verification via Tool-Integrated Reinforcement Learning

Large language models have achieved strong performance on medical reasoning benchmarks, yet their deployment in clinical settings demands rigorous verification to ensure factual accuracy. While reward models offer a scalable approach for reasoning trace verification, existing methods face two limitations: they produce only scalar reward values without explicit justification, and they rely on single-pass retrieval that precludes adaptive knowledge access as verification unfolds. We introduce $\method$, an agentic framework that addresses these limitations by training medical reasoning verifiers to iteratively query external medical corpora during evaluation. Our approach combines tool-augmented verification with an iterative reinforcement learning paradigm that requires only trace-level supervision, alongside an adaptive curriculum mechanism that dynamically adjusts training data distribution. Across four medical reasoning benchmarks, $\method$ achieves substantial gains over existing methods, improving MedQA accuracy by 23.5% and MedXpertQA by 32.0% relative to the base generator in particular. Crucially, $\method$ demonstrates an $\mathbf{8\times}$ reduction in sampling budget requirement compared to prior reward model baselines. These findings establish that grounding verification in dynamically retrieved evidence offers a principled path toward more reliable medical reasoning systems.

cs.AI↗
arXiv2026

MRAG-Suite: A Diagnostic Evaluation Platform for Visual Retrieval-Augmented Generation

Multimodal Retrieval-Augmented Generation (Visual RAG) significantly advances question answering by integrating visual and textual evidence. Yet, current evaluations fail to systematically account for query difficulty and ambiguity. We propose MRAG-Suite, a diagnostic evaluation platform integrating diverse multimodal benchmarks (WebQA, Chart-RAG, Visual-RAG, MRAG-Bench). We introduce difficulty-based and ambiguity-aware filtering strategies, alongside MM-RAGChecker, a claim-level diagnostic tool. Our results demonstrate substantial accuracy reductions under difficult and ambiguous queries, highlighting prevalent hallucinations. MM-RAGChecker effectively diagnoses these issues, guiding future improvements in Visual RAG systems.

cs.CL↗
arXiv2026

Retrieval--Reasoning Processes for Multi-hop Question Answering: A Four-Axis Design Framework and Empirical Trends

Multi-hop question answering (QA) requires systems to iteratively retrieve evidence and reason across multiple hops. While recent RAG and agentic methods report strong results, the underlying retrieval--reasoning \emph{process} is often left implicit, making procedural choices hard to compare across model families. This survey takes the execution procedure as the unit of analysis and introduces a four-axis framework covering (A) overall execution plan, (B) index structure, (C) next-step control (strategies and triggers), and (D) stop/continue criteria. Using this schema, we map representative multi-hop QA systems and synthesize reported ablations and tendencies on standard benchmarks (e.g., HotpotQA, 2WikiMultiHopQA, MuSiQue), highlighting recurring trade-offs among effectiveness, efficiency, and evidence faithfulness. We conclude with open challenges for retrieval--reasoning agents, including structure-aware planning, transferable control policies, and robust stopping under distribution shift.

cs.CL↗
arXiv2025

Orchestrator Multi-Agent Clinical Decision Support System for Secondary Headache Diagnosis in Primary Care

Unlike most primary headaches, secondary headaches need specialized care and can have devastating consequences if not treated promptly. Clinical guidelines highlight several 'red flag' features, such as thunderclap onset, meningismus, papilledema, focal neurologic deficits, signs of temporal arteritis, systemic illness, and the 'worst headache of their life' presentation. Despite these guidelines, determining which patients require urgent evaluation remains challenging in primary care settings. Clinicians often work with limited time, incomplete information, and diverse symptom presentations, which can lead to under-recognition and inappropriate care. We present a large language model (LLM)-based multi-agent clinical decision support system built on an orchestrator-specialist architecture, designed to perform explicit and interpretable secondary headache diagnosis from free-text clinical vignettes. The multi-agent system decomposes diagnosis into seven domain-specialized agents, each producing a structured and evidence-grounded rationale, while a central orchestrator performs task decomposition and coordinates agent routing. We evaluated the multi-agent system using 90 expert-validated secondary headache cases and compared its performance with a single-LLM baseline across two prompting strategies: question-based prompting (QPrompt) and clinical practice guideline-based prompting (GPrompt). We tested five open-source LLMs (Qwen-30B, GPT-OSS-20B, Qwen-14B, Qwen-8B, and Llama-3.1-8B), and found that the orchestrated multi-agent system with GPrompt consistently achieved the highest F1 scores, with larger gains in smaller models. These findings demonstrate that structured multi-agent reasoning improves accuracy beyond prompt engineering alone and offers a transparent, clinically aligned approach for explainable decision support in secondary headache diagnosis.

cs.AI↗
arXiv2025

Generative Foundation Model for Structured and Unstructured Electronic Health Records

Electronic health records (EHRs) are rich clinical data sources but complex repositories of patient data, spanning structured elements (demographics, vitals, lab results, codes), unstructured clinical notes and other modalities of data. Harnessing this heterogeneity is critical for improving patient outcomes. Recent advances in large language models (LLMs) have enabled foundation models that can learn from multiple data modalities and support clinical tasks. However, most current approaches simply serialize numeric EHR data into text, which risks losing temporal and quantitative detail. We introduce Generative Deep Patient (GDP), a multimodal foundation model that natively encodes structured EHR time-series via a CNN-Transformer encoder and fuses it with unstructured EHRs through cross-modal attention into a LLaMA-based decoder. GDP is trained in two stages: (1) generative pretraining, where it learns to produce clinical narratives from raw patient timelines while also performing masked feature prediction (MFP) and next time-step prediction (NTP) to capture temporal dynamics; and (2) multi-task fine-tuning for clinically meaningful predictions (e.g., heart failure, type 2 diabetes, 30-day readmission). In clinical prediction, GDP demonstrated superior performance on MIMIC-IV: heart failure AUROC = 0.923, type 2 diabetes AUROC = 0.817, and 30-day readmission AUROC = 0.627. For narrative generation, GDP achieved ROUGE-L = 0.135 and BERTScore-F1 = 0.545. In a blinded human evaluation, GDP-Instruct scored highest on faithfulness, fluency, and overall clinical utility, suggesting reduced hospital documentation workload without sacrificing accuracy. Our results demonstrate that a single multimodal foundation model can both predict clinically actionable events and generate high-quality clinical narratives. Furthermore, GDP's flexible architecture can be extended to additional modalities.

cs.AI↗
arXiv2025

DeepRAG: Integrating Hierarchical Reasoning and Process Supervision for Biomedical Multi-Hop QA

We propose DeepRAG, a novel framework that integrates DeepSeek hierarchical question decomposition capabilities with RAG Gym unified retrieval-augmented generation optimization using process level supervision. Targeting the challenging MedHopQA biomedical question answering task, DeepRAG systematically decomposes complex queries into precise sub-queries and employs concept level reward signals informed by the UMLS ontology to enhance biomedical accuracy. Preliminary evaluations on the MedHopQA dataset indicate that DeepRAG significantly outperforms baseline models, including standalone DeepSeek and RAG Gym, achieving notable improvements in both Exact Match and concept level accuracy.

cs.CL↗
arXiv2025

Curriculum Guided Reinforcement Learning for Efficient Multi Hop Retrieval Augmented Generation

Retrieval-augmented generation (RAG) grounds large language models (LLMs) in up-to-date external evidence, yet existing multi-hop RAG pipelines still issue redundant subqueries, explore too shallowly, or wander through overly long search chains. We introduce EVO-RAG, a curriculum-guided reinforcement learning framework that evolves a query-rewriting agent from broad early-stage exploration to concise late-stage refinement. EVO-RAG couples a seven-factor, step-level reward vector (covering relevance, redundancy, efficiency, and answer correctness) with a time-varying scheduler that reweights these signals as the episode unfolds. The agent is trained with Direct Preference Optimization over a multi-head reward model, enabling it to learn when to search, backtrack, answer, or refuse. Across four multi-hop QA benchmarks (HotpotQA, 2WikiMultiHopQA, MuSiQue, and Bamboogle), EVO-RAG boosts Exact Match by up to 4.6 points over strong RAG baselines while trimming average retrieval depth by 15 %. Ablation studies confirm the complementary roles of curriculum staging and dynamic reward scheduling. EVO-RAG thus offers a general recipe for building reliable, cost-effective multi-hop RAG systems.

cs.CL↗
arXiv2025

Learning from Committee: Reasoning Distillation from a Mixture of Teachers with Peer-Review

While reasoning capabilities typically emerge in large language models (LLMs) with tens of billions of parameters, recent research focuses on improving smaller open-source models through knowledge distillation (KD) from commercial LLMs. However, many of these studies rely solely on responses from a single LLM as the gold rationale, unlike the natural human learning process, which involves understanding both the correct answers and the reasons behind mistakes. In this paper, we introduce a novel Fault-Aware DistIllation via Peer-Review (FAIR) approach: 1) instead of merely obtaining rationales from teachers, our method asks teachers to identify and explain the student's mistakes, providing customized instruction learning data; 2) we design a simulated peer-review process between teacher LLMs, and selects only the generated rationales above the acceptance threshold, which reduces the chance of teachers guessing correctly with flawed rationale, improving instructional data quality. Comprehensive experiments and analysis on mathematical, commonsense, and logical reasoning tasks demonstrate the effectiveness of our method. Our code is available at https://github.com/zhuochunli/Learn-from-Committee.

cs.CL↗
arXiv2025

ReasoningRank: Teaching Student Models to Rank through Reasoning-Based Knowledge Distillation

Reranking documents based on their relevance to a given query is a critical task in information retrieval. Traditional reranking methods often lack transparency and rely on proprietary models, hindering reproducibility and interpretability. We propose Reason-to-Rank (R2R), a novel open-source reranking approach that enhances transparency by generating two types of reasoning: direct relevance reasoning, which explains how a document addresses the query, and comparison reasoning, which justifies the relevance of one document over another. We leverage large language models (LLMs) as teacher models to generate these explanations and distill this knowledge into smaller, openly available student models. Our student models are trained to generate meaningful reasoning and rerank documents, achieving competitive performance across multiple datasets, including MSMARCO and BRIGHT. Experiments demonstrate that R2R not only improves reranking accuracy but also provides valuable insights into the decision-making process. By offering a structured and interpretable solution with openly accessible resources, R2R aims to bridge the gap between effectiveness and transparency in information retrieval, fostering reproducibility and further research in the field.

cs.CL↗
arXiv2025

Memory-Aware and Uncertainty-Guided Retrieval for Multi-Hop Question Answering

Multi-hop question answering (QA) requires models to retrieve and reason over multiple pieces of evidence. While Retrieval-Augmented Generation (RAG) has made progress in this area, existing methods often suffer from two key limitations: (1) fixed or overly frequent retrieval steps, and (2) ineffective use of previously retrieved knowledge. We propose MIND (Memory-Informed and INteractive Dynamic RAG), a framework that addresses these challenges through: (i) prompt-based entity extraction to identify reasoning-relevant elements, (ii) dynamic retrieval triggering based on token-level entropy and attention signals, and (iii) memory-aware filtering, which stores high-confidence facts across reasoning steps to enable consistent multi-hop generation.

cs.CL↗
arXiv2025

Bias Evaluation and Mitigation in Retrieval-Augmented Medical Question-Answering Systems

Medical Question Answering systems based on Retrieval Augmented Generation is promising for clinical decision support because they can integrate external knowledge, thus reducing inaccuracies inherent in standalone large language models (LLMs). However, these systems may unintentionally propagate or amplify biases associated with sensitive demographic attributes like race, gender, and socioeconomic factors. This study systematically evaluates demographic biases within medical RAG pipelines across multiple QA benchmarks, including MedQA, MedMCQA, MMLU, and EquityMedQA. We quantify disparities in retrieval consistency and answer correctness by generating and analyzing queries sensitive to demographic variations. We further implement and compare several bias mitigation strategies to address identified biases, including Chain of Thought reasoning, Counterfactual filtering, Adversarial prompt refinement, and Majority Vote aggregation. Experimental results reveal significant demographic disparities, highlighting that Majority Vote aggregation notably improves accuracy and fairness metrics. Our findings underscore the critical need for explicitly fairness-aware retrieval methods and prompt engineering strategies to develop truly equitable medical QA systems.

cs.CL↗