arXiv subjects

Juntao Li

Publications and source records attributed to Juntao Li.

At least 19 recordsLinked to original sources

arXiv2026

Prism Attention: Proposal-Refined Index Sharing Mechanism for Efficient LLMs Inference

Long-context decoding in large language models (LLMs) is increasingly bottlenecked by the memory and computation required to attend over an ever-growing Key-Value (KV) cache at each autoregressive step. Although token-level sparse attention reduces this burden by restricting computation to a small subset of relevant past tokens, its practical gains are often offset by repeatedly recomputing layer-wise sparse indices as the context grows. A common remedy is to share sparse indices across layers by computing them only at designated anchor layers, but such static reuse ignores the layer-wise evolution of attention patterns, causing the reused indices to progressively deviate from the layer-specific optima and incur substantial attention loss. By revisiting cross-layer sparsity from the perspective of attention dynamics, we find that inter-layer salient token shift is highly localized: although the exact top-$κ$ indices differ across layers, the core high-attention tokens remain stable, with changes concentrated near the expanded boundary of previously selected regions. Motivated by this, we propose Prism Attention, a proposal-refined sparse attention framework. It replaces rigid cross-layer index reuse with a coarse-to-fine strategy, preserving layer-wise adaptability while retaining the efficiency benefits of cross-layer sharing. Extensive experiments show that Prism Attention achieves state-of-the-art performance among mainstream sparse attention methods on both reasoning and long-context benchmarks, while attaining a $3.4\times$ end-to-end decoding speedup over FlashAttention at 128K context length.

cs.CL↗

arXiv2026

EviRerank: Adaptive Evidence Construction for Long-Document LLM Reranking

Decoder-only LLM rerankers struggle with long documents: inference is costly and relevance signals can be diluted by irrelevant context. Motivated by an attention analysis indicating a consistent degradation trend when non-relevant text is appended, we propose EviRerank, an evidence-based long-document reranking framework for decoder-only LLMs. EviRerank (i) scores document blocks with a lightweight selector (BM25, bi-encoder, or cross-encoder), (ii) constructs a compact reranking context under a hard token cap by dynamically budgeting evidence blocks with Adaptive Evidence Budgeting (AEB) and adding a global summary cue via Summary Augmentation (SA), and (iii) reranks with a decoder-only LLM. Across TREC DL'19, DL'22, DL'23, and MLDR-zh, EviRerank consistently outperforms full-document LLM reranking and strong block-selection baselines while substantially reducing the required input length. On TREC DL'19, EviRerank reaches up to 0.744 nDCG@10 while substantially reducing reranking latency. Additional RankZephyr-7B experiments confirm transfer to listwise reranking under matched context budgets.

cs.IR↗

arXiv2026

OmniSelect: Dynamic Modality-Aware Token Compression for Efficient Omni-modal Large Language Models

Omnimodal large language models (OmniLLMs) have recently gained increasing attention for unified audio-video understanding. However, processing long multimodal token sequences introduces substantial computational overhead, making efficient token compression crucial. Existing methods typically rely on fixed, modality-specific guidance, which fails to account for the varying importance of modalities across different queries. To address this limitation, we propose $\textbf{OmniSelect}$, a training-free, modality-adaptive token pruning framework that dynamically selects appropriate compression strategies for multimodal inputs. Specifically, we leverage a lightweight AudioCLIP model to estimate cross-modal relevance and categorize each input into three pruning regimes: Audio-Centric, Video-Centric, and Uniform pruning. Based on these relevance scores, OmniSelect further performs fine-grained token pruning within each temporal group, adaptively allocating pruning ratios to preserve informative tokens across modalities. By explicitly modeling modality preference and enabling dynamic strategy selection, OmniSelect effectively avoids the pitfalls of one-size-fits-all compression. Extensive experiments demonstrate that our method achieves efficient multimodal token reduction while maintaining strong performance, without requiring any additional training.

cs.CV↗

arXiv2026

MemReread: Enhancing Agentic Long-Context Reasoning via Memory-Guided Rereading

Long-context reasoning tasks have motivated agent-memory approaches that process document chunks sequentially while maintaining a dynamically updated memory, avoiding the quadratic complexity of standard attention. Existing methods often rely on retrieval to recover information discarded during memory overwriting, but remain vulnerable to evidence loss during memory formation and interference from invalid queries. To overcome these limitations, we propose MemReread. Built upon streaming reading, it circumvents intermediate retrieval, triggering question decomposition and targeted rereading when the final memory is insufficient. By recovering indirect evidence that may have been prematurely discarded, MemReread enables non-linear reasoning while preserving the logical flow of document comprehension. We further introduce a reinforcement learning framework that improves length extrapolation and adaptively determines the number of rereading passes according to task complexity, enabling flexible control of computational overhead. Our MemReread-4B outperforms strong baselines on long-context reasoning tasks (e.g., +1.6pp v.s. DeepSeek-V4-Pro on 2WikiMultiHopQA) while maintaining linear time complexity with respect to context length.

cs.CL↗

arXiv2026

Judging in Latent Space: Efficient Generative Reward Modeling via Semantics-Preserving Compression

Reward modeling often requires jointly representing and reasoning over multiple evaluation criteria, yet verbalizing this process token by token can incur substantial inference cost. Recent work on latent reasoning suggests that continuous states may support this computation more compactly. We introduce LatentGRM, a latent evaluation framework built on semantic chunking, compression, and reconstruction. By using the structure of rubric-guided evaluations to guide compression, LatentGRM learns compact continuous trajectories that support autonomous pairwise judgments without generating textual assessments. A separate interpreter reconstructs evaluation text from these trajectories, providing an offline view of the information retained under compression. Under matched training data and backbones, LatentGRM achieves competitive aggregate preference accuracy relative to explicit Supervised Fine-Tuning (SFT) judges at both 4B and 8B scales. Across four benchmark domains, LatentGRM-8B compresses evaluation trajectories by 8.9--9.2x and reduces total judge inference time by 6.1--7.0x at vote@5. Controlled rubric interventions show that criterion-dependent preference information is carried through the latent sequence. Together, these results demonstrate that continuous latent evaluation can substantially reduce inference cost while preserving competitive judgment quality.

cs.CL↗

arXiv2026

MMLongCite: A Benchmark for Evaluating Faithfulness of Long-Context Vision-Language Models

The rapid advancement of long-context vision language models (LCVLMs) has led to a significant expansion of their context windows. However, an extended context window does not guarantee the effective utilization of the context, posing a critical challenge for real-world applications. Current evaluations of such long-context faithfulness in multimodal settings remain limited to short contexts. To bridge this gap, we introduce MMLongCite, the first benchmark evaluating the faithfulness of LCVLMs via multimodal citation generation. MMLongCite features 2,280 examples across 8 tasks and diverse modalities (image, video, interleaved), with context lengths scaled from 16K to 128K tokens. To test spatial localization capabilities of LCVLMs, we also introduce MMLongCite-HR, evaluating fine-grained visual grounding amidst dense pixel spaces. Through extensive benchmarking of cutting-edge LCVLMs, we provide a systematic analysis of current multimodal citation capabilities. Our results reveal a significant discrepancy between answer correctness and citation faithfulness. We also conduct attention pattern investigations and in-depth error analyses to reveal the underlying phenomena of failures in LCVLMs. MMLongCite establishes a rigorous foundation for diagnosing and advancing the faithfulness of LCVLMs. We hope our findings provide meaningful insights to drive further improvements in the long-context capabilities of LCVLMs.

cs.CV↗

arXiv2026

Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference

The quadratic computational complexity of standard attention mechanisms presents a severe scalability bottleneck for LLMs in long-context scenarios. While hybrid attention mechanisms combining Full Attention (FA) and Sparse Attention (SA) offer a potential solution, existing methods typically rely on static allocation ratios that fail to accommodate the variable retrieval demands of different tasks. Furthermore, head-level dynamic sparsity often introduces severe computational load imbalance and synchronization long-tails, which hinder hardware acceleration during autoregressive decoding. To bridge this gap, we introduce Flux Attention, a context-aware framework that dynamically optimizes attention computation at the layer level. By integrating a lightweight Layer Router into frozen pretrained LLMs, the proposed method adaptively routes each layer to FA or SA based on the input context. This layer-wise routing preserves high-fidelity information retrieval while ensuring contiguous memory access, translating theoretical computational reductions into practical wall-clock speedups. As a parameter-efficient approach, our framework requires only 12 hours of training on 8$\times$A800 GPUs. Extensive experiments across multiple long-context and mathematical reasoning benchmarks demonstrate that Flux Attention achieves a superior trade-off between performance and inference speed compared with baseline models, with speed improvements of up to $2.8\times$ and $2.0\times$ in the prefill and decode stages.

cs.LG↗

arXiv2026

GraspTune: Tactile-Driven Execution Refinement for Robust Grasping

Visual grasp proposal generation has advanced rapidly, yet converting a selected proposal into a stable physical grasp remains a central execution-stage challenge. This paper introduces GraspTune, a tactile-driven execution-stage refinement framework that starts from a nominal proposal and applies bounded residual TCP motions during approach, contact formation, and final grasp execution. GraspTune learns control-facing contact semantics from local depth, tactile signals, state, and history using state-conditioned expert contact queries and multi-task supervision for contact change, contact risk, and post-close readiness. The representation conditions a diffusion-pretrained residual policy and is aligned with PPO for closed-loop execution. Across more than 60,000 simulated executions over 20 object categories, GraspTune establishes an execution-layer benefit across four proposal generators, raising stable grasp success by +19.22, +9.55, +12.45, and +20.70 percentage points for GraspNet, Contact-GraspNet, AnyGrasp, and VGN. A four-fold held-out category study raises unseen-object execution from 54.58% to 70.33%, showing category-disjoint generalization of contact correction. Across more than 1,000 real-robot trials on a UR5e setup with Xense fingertip sensors, GraspTune raises GraspNet execution from 71.0% to 84.3%, validating direct transfer without realworld policy fine-tuning. Together, these results turn visually plausible proposals into stable physical grasps for downstream contact-rich manipulation. A supplementary video is available at https://youtu.be/kcq7fSLNtzU.

cs.RO↗

arXiv2026

D$^3$-MOPD: Dynamic Domain ScheDuling for Efficient Multi-Teacher Distillation

Multi-teacher on-policy distillation (MOPD) distills several domain-expert teachers into a single student by minimizing per-domain reverse-KL divergence on the student's own rollouts. Existing approaches typically fix the per-domain data mixture before training, overlooking the fact that different domains converge at substantially different rates: some plateau early while others continue to improve throughout the training budget. A fixed mixture therefore wastes compute on fast-converging domains and undertrains slower-converging ones. To address this, we propose D$^3$-MOPD (Dynamic Domain ScheDuling for MOPD), a zero-overhead scheduler that repurposes the per-domain reverse-KL signal already produced during training to adapt the domain mixture online. Running asynchronously outside the training process, an off-process watcher periodically tracks each domain's KL trajectory, estimates remaining headroom and current improvement rate, and accordingly adjusts the domain sampling ratios without altering the core training loop. Our D$^3$-MOPD scales naturally to arbitrary numbers of domains, and the expected benefit grows as more domains introduce more diverse convergence patterns for the scheduler to exploit. On a Qwen3.6-35B-A3B student distilled from four domain-expert teachers, D$^3$-MOPD closes 97% of the average student-to-teacher performance gap, compared with 63% for vanilla MOPD, reaches the same peak performance with an approximately 3$\times$ reduction in rollout steps, and surpasses the specialist teachers on three of seven benchmarks.

cs.LG↗

arXiv2026

Agent-as-Peer-Debriefer: A Multi-Agent Framework with Perspective-Based Refinement for Qualitative Analysis

Large language models (LLMs) are increasingly used for qualitative data analysis (QDA), yet their outputs often miss the depth and nuance of human analysis. We argue this gap reflects a missing credibility practice from human QDA: peer debriefing, in which an analyst seeks feedback from a disinterested peer and uses it to refine their coding. To bring this practice into LLM-assisted QDA, we propose Agent-as-Peer-Debriefer, a multi-agent QDA framework that builds peer debriefing into key coding steps. In our framework, a Hierarchical Coding Agent follows the standard QDA process to generate codes, sub-themes, and themes, along with self-explanations and reflection memos. It then shares these outputs with three Peer-Debriefing Agents, each applying a distinct analytical perspective (Theory-Driven, Data-Driven, or Applied) and refining the codes by keeping, renaming, reassigning, merging, or splitting them. These perspectives are drawn from established human QDA practices that generalize across domains and datasets. To evaluate the framework, we test it on three datasets across two domains with three LLMs, measuring semantic similarity to human-annotated codes. Across all settings, perspective-based, peer-debriefing refinement aligns more closely with human codes than a single-LLM baseline, and an ablation further shows the gain is not merely from additional refinement. The three perspectives also produce distinct trade-offs, showing that the choice of perspective is a meaningful and controllable design decision. More broadly, these findings suggest that simulating peer debriefing with explicit perspectives is a promising route to more credible LLM-assisted QDA.

cs.AI↗

arXiv2026

MMLongEmbed: Benchmarking Multimodal Embedding Models in Long-Context Scenarios

Recent advancements have significantly expanded the theoretical context windows of Multimodal Embedding Models (MEMs). However, larger context windows do not necessarily translate into effective comprehension and representation of long-context multimodal inputs, which remains a critical bottleneck for real-world deployment. To address the lack of systematic evaluation in this setting, we introduce MMLongEmbed, the first comprehensive benchmark for evaluating MEMs in long-context scenarios. MMLongEmbed comprises four retrieval tasks spanning multiple context-length ranges, covering text, document, and video modalities. Through extensive evaluation of state-of-the-art models, we find that current architectures rely heavily on superficial feature matching and struggle to capture deep semantic and structural dependencies. We further observe that performance degradation varies systematically with context length and key information placement. Moreover, models exhibit substantially different robustness to redundant contextual information across modalities. For reproducibility, the benchmark and code are publicly available.

cs.CV↗

arXiv2026

SABER: Stability-Aware Early Exit for LLM Reasoning via Adversarial Branch Probing

Large Reasoning Models (LRMs) achieve strong reasoning capabilities, yet long-chain reasoning becomes inefficient once the intermediate answer stabilizes across reasoning steps: additional reasoning yields little marginal benefit while incurring substantial inference cost. Existing early-exit methods based on confidence or entropy poorly capture reasoning stability, while consistency-based approaches rely on multi-step trajectory agreement, requiring sequential evaluations that delay exit. To better balance efficiency and reliability, we propose SABER, a training-free framework for stability-aware early exit via adversarial branch probing. SABER constructs simple yet effective semantic perturbations around intermediate reasoning states to form adversarial branches, and applies lightweight probing to estimate their likely final outcomes without full trajectory rollouts. When the probed outcomes remain consistent across branches, SABER exits early; otherwise, it continues reasoning. Experiments across multiple reasoning benchmarks and model architectures show that SABER reduces reasoning token consumption by 30.2\%--39.8\% on average while maintaining competitive accuracy with full-length reasoning.

cs.AI↗

arXiv2026

MediRec: Enhancing Chinese Medication Recommendation with Explainable Clinical Reasoning

Large language models (LLMs) have shown strong potential for clinical decision support through their advanced language understanding and reasoning capabilities. However, their application to Chinese clinical medication recommendation remains largely unexplored. Existing approaches are primarily developed on English electronic health record datasets and focus on coarse-grained medication code prediction, offering limited support for interpretable clinical decision-making. In this work, we propose MediRec, an explainable LLM-based framework for Chinese medication recommendation from electronic health records. MediRec combines clinically grounded reasoning-chain distillation with reinforcement learning to improve both recommendation accuracy and interpretability. Comprehensive experiments on a Chinese medication recommendation benchmark show that MediRec achieves strong performance, with an F1 score of 0.5813 and a Jaccard score of 0.4626. Further analyses indicate that MediRec generates clinically plausible recommendations with transparent reasoning, demonstrating its effectiveness for explainable medication decision support in Chinese healthcare settings.

cs.CL↗

arXiv2026

High resolution meta-stereomicroscope based on birefringent meta-optics

Achieving both high lateral and high depth resolution is a longstanding goal in stereomicroscopy. Although meta-optics have revolutionized lens design by alleviating the physical constraints of conventional optical architectures, existing metalens-assisted stereomicroscopes still suffer from field of view (FOV) mismatch between meta-optical and conventional optical components in stereomicroscopes, thereby limiting their imaging performance. Here, we show that this mismatch can be fundamentally eliminated through a fully meta-optical architecture. The integrated system enables flexible control of numerical aperture and magnification with larger depth of field (DOF) and FOV than those of the metalens-assisted stereomicroscopes. Experimentally, we achieve an integrated meta-stereomicroscope with a lateral resolution of 435 nm, surpassing the performance of previously reported stereomicroscopes. Empowered by a stereo neural network, the system enables straightforward reconstruction of high-resolution three-dimensional surface morphology with a depth resolution of 1026 nm, demonstrating the capability to simultaneously achieve high lateral and depth resolution imaging. This integrated architecture operates in both transmission and reflection modes for biomedical imaging and industrial inspection, highlighting its broad applicability for real-time observation across biomedical and industrial scenarios.

physics.optics↗

arXiv2026

When to Vote, When to Rewrite: Disagreement-Guided Strategy Routing for Test-Time Scaling

Large Reasoning Models (LRMs) achieve strong performance on mathematical reasoning tasks but remain unreliable on challenging instances. Existing test-time scaling methods, such as repeated sampling, self-correction, and tree search, improve performance at the cost of increased computation, yet often exhibit diminishing returns on hard problems. We observe that output disagreement is strongly correlated with instance difficulty and prediction correctness, providing a useful signal for guiding instance-level strategy selection at test time. Based on this insight, we propose a training-free framework that formulates test-time scaling as an instance-level routing problem, rather than allocating more computation within a single strategy, dynamically selecting among different scaling strategies based on output disagreement. The framework applies lightweight resolution for consistent cases, majority voting for moderate disagreement, and rewriting-based reformulation for highly ambiguous instances. Experiments on seven mathematical benchmarks and three models show that our method improves accuracy by 3% - 7% while reducing sampling cost compared to existing approaches.

cs.AI↗

arXiv2026

ProgramTab: Boosting Table Reasoning of LLMs via Programmatic Paradigm

Table-based reasoning with large language models (LLMs), which requires reasoning based on natural language questions and structured tabular data, has gained widespread attention. However, a series of issues still constrain the application of this task. The previous approaches suffered from significant performance degradation when faced with large tables due to the difficulty of long text modeling and the limitation of input length for LLMs. The text-to-SQL approach is used to efficiently extract key information from tables and generate smaller sub-tables. However, tabular data, especially web tables, often lack the necessary structure and consistency, making them unsuitable for performing mathematical logic operations using SQL queries. We propose the ProgramTab framework, which guides LLMs employing in-context learning to perform tabular data preprocessing with Python code, as well as the momentous contents extraction with row and column extraction and SQL generation. The experiment results on table reasoning datasets demonstrate that the ProgramTab framework effectively deals with table-based reasoning tasks and outperforms all LLM-based baselines.

cs.CL↗

arXiv2026

Homer: Understanding Long-form Videos with Hierarchical Memory and Agentic Reasoning

Multimodal large language models excel on short clips but struggle on hour-long videos in an online setting, where frames are processed incrementally under limited memory. Existing online methods either retain compact visual representations that lack semantic structure, or build higher-level memory stores organized around temporal proximity rather than explicit causal links, leaving multi-hop narrative reasoning to be reconstructed by the LLM at every query. We bridge this gap with \textsc{Homer}, a Hierarchical Online Memory Exploration and Reasoning framework. \textsc{Homer}'s memory mirrors the multi-scale structure of long videos, ranging from raw perception, to recurring entities, to events connected by explicit temporal and causal relations. Its agentic reasoner then explores this memory the way humans do, locating the relevant scene, looking up details, and composing the answer through multi-round memory retrieval, with a harness that verifies and corrects each step. \textsc{Homer} outperforms the previous best agent method by $+5.5$, $+10.8$, and $+4.4$ points on M3-Bench-robot, M3-Bench-web, and Video-MME-Long, and consistently lifts three various LLM backbones, indicating a model-agnostic structural capability for grounded retrieval over long videos.

cs.CV↗

arXiv2026

IS-CoT: Breaking the Long-form Generation Collapse via Interleaved Structural Thinking

Generating coherent and controllable long-form content remains a persistent challenge for Large Language Models (LLMs). While reasoning-enhanced models have demonstrated success in logic-intensive domains, our evaluation reveals that they suffer from a severe length collapse in open-ended writing, where performance degrades sharply as target lengths exceed 2,000 words. We attribute this failure to the limitation of static hierarchical planning, which struggles to provide dynamic guidance over extended contexts. To bridge this gap, we introduce the Interleaved Structural Chain-of-Thought (IS-CoT) framework. Unlike external agentic workflows, IS-CoT embeds a dynamic Plan-Write-Reflect cycle into the generation process, enabling continuous strategy adaptation and global alignment without additional assistance. Based on this framework, we construct a high-quality dataset of interleaved reasoning traces via a multi-teacher pipeline and train IS-Writer-8B. Experiments demonstrate that IS-Writer-8B achieves state-of-the-art performance on challenging long-form benchmarks (e.g., +3.08 vs. DeepSeek-V3.2 on LongBench-Write), exhibiting robust length compliance and coherence competitive with significantly larger proprietary models.

cs.CL↗