Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.CV) 2026-06-25

Disease-Centric Vision-Language Pretraining with Hybrid Visual Encoding for 3D Computed Tomography

Vision-language pre-training (VLP) holds great promise for general-purpose medical AI by leveraging radiology reports as rich textual supervision, yet existing methods struggle with 3D CT imaging due to inefficient visual backbones and coarse semantic alignment. To address these issues, we propose a tailored VLP framework featuring three key components: (1) a CNN-ViT hybrid encoder that replaces ViT's patch embedding with a 3D CNN backbone to efficiently capture local anatomical details while preserving global attention and compatibility with pre-trained cross-modal priors; (2) a disease-level contrastive learning mechanism using learnable query tokens to dynamically extract disease-specific semantics from full reports and align them with corresponding visual features, thereby disentangling distinct diseases within the same anatomical region; and (3) a diagnosis-aware prompt strategy that employs real clinical phrases and aggregated disease prototypes to bridge the pre-training-inference gap and enhance zero-shot diagnostic reliability. Our model achieves state-of-the-art performance on CT-RATE (84.4% AUC, +5.1%) and Rad-ChestCT (75.4% AUC, +5.4%), with even larger gains (+9.8% AUC) on a challenging 60-disease benchmark, and demonstrates strong transferability to radiology report generation, underscoring the generality and clinical utility of our approach.

02.
arXiv (CS.CL) 2026-06-16

Retrievable Gradients: Continual Post-Training Without Cumulative Weight Drift

Continual post-training enables models to absorb emerging knowledge after deployment, but repeatedly updating shared parameters can accumulate weight drift, potentially causing catastrophic forgetting and degrading general capabilities. Retrieval-augmented generation avoids such parameter drift, yet often lacks the depth of parametric knowledge integration. In this paper, we propose ReGrad (Retrievable Gradients), a new paradigm that treats gradients as retrievable units of knowledge. ReGrad pre-computes document-specific gradients offline, stores them in an indexed Gradient Bank, and retrieves only query-relevant gradients at inference time for temporary weight adaptation. However, raw language-modeling gradients are optimized for token-level document reconstruction rather than for query-driven knowledge use. We therefore introduce a bi-level meta-learning objective that reshapes document-derived gradients into generalizable adaptation signals for downstream tasks. Experiments across general and domain-specific settings show that \textsc{ReGrad} outperforms CPT and RAG baselines, enabling scalable and reversible parametric knowledge injection without accumulating weight drift.

03.
arXiv (CS.LG) 2026-06-15

NeST: Neuron Selective Tuning for LLM Safety

arXiv:2602.16835v2 Announce Type: replace-cross Abstract: Safety alignment is essential for the responsible deployment of Large Language Models (LLMs). Yet, existing approaches often rely on heavyweight fine-tuning that is costly to update, audit, and maintain across model families. Full fine-tuning incurs substantial computational and storage overhead, while parameter-efficient methods, e.g., Low-Rank Adaptation (LoRA), trade efficiency for inconsistent safety gains and sensitivity to design choices. Safety intervention mechanisms reduce unsafe outputs without modifying model weights, but do not directly shape or preserve the internal representations that govern safety behavior. We present NeST, a Neuron-Selective Tuning framework for efficient post-hoc safety alignment. NeST identifies safety-relevant feed-forward neurons via activation probing on vanilla harmful and benign prompts, clusters neurons with similar activation profiles, and trains shared cluster-level updates while freezing the rest of the model. Importantly, NeST is trained only on vanilla malicious prompts, without using jailbreak-specific attack data, yet generalizes robustly to diverse jailbreaks. The learned updates are then folded into the original weights, incurring no inference-time overhead. Evaluated on 14 open-weight language and multimodal models, NeST outperforms lightweight baselines and approaches full fine-tuning robustness with significantly fewer trainable parameters. On text-only models, NeST reduces average jailbreak attack success rate from 44.5% to 1.1% while training only 0.4M parameters on average. Across multimodal settings, it reduces ASR from 55.3% to 1.1%, and for downstream fine-tuned variants, it restores safety by reducing ASR from 53.8% to 0.8%. These results show that robust, maintainable safety alignment can be achieved by concentrating adaptation on localized, functionally coherent safety structures.

04.
arXiv (CS.CV) 2026-06-18

Where Will They Go? Modelling Multimodal Pedestrian Manoeuvres from Ego-centric Videos

Pedestrian trajectory prediction from an ego-centric camera is challenging since it depends on complex interactions with vehicles and scene context, as well as the intention of the pedestrian. By modelling correlation and intent from the historical and future trajectories of the pedestrian, it will usually result in a multimodal (i.e. multiple modes) distribution. Existing stochastic predictors often sample multiple futures from a single unimodal distribution, which can yield sub-optimal 'mixed-mode' trajectories that lie between distinct motion patterns and become implausible in real scenes. In this paper, we propose MMPM, a mode-aware framework that separately models future trajectory distributions into semantically meaningful modes based on the pedestrian's crossing behavior. MMPM consists of two modules: behavior-aware Pedestrian Interaction Module (PIM) that jointly captures pedestrian-vehicle and pedestrian-environment interactions by introducing gaze, head and hand gesture, and a CVAE-based Mode-aware Trajectory Predictor (MTP) module to model the future trajectory distributions on two modes, crossing and non-crossing the road, separately. A query-based decoder further enforces mode consistency during decoding. Experiments on PIE and JAAD datasets show that our method surpasses state-of-the-art baselines. Our proposed MTP is model-agnostic, which can be integrated into existing frameworks such as BiTrap-NP and SGNet-ED to further improve future trajectory prediction performance. We additionally introduce a data-driven validation protocol that matches predictions to spatio-temporally consistent ground-truth trajectories, demonstrating improved frame-wise displacement errors over previous work.

05.
arXiv (CS.CL) 2026-06-16

SHARD: Safe and Helpful Alignment via Self-Reframing Distillation

Large language models often struggle with sensitive prompts. They may refuse outright, provide generic safety boilerplate, or fail to address the user's legitimate informational needs that can be answered safely. We introduce SHARD, a self-reframing distillation method to improve safe-helpfulness. It first rewrites sensitive prompts to surface benign intent using philosophical guidelines, then reframes its original responses into safe, more helpful ones, and finally fine-tunes the model on its self-reframed responses. Across DNA and the English subset of LINGUASAFE, SHARD improves helpfulness for most model families while preserving safety. It also remains competitive with distillation from a larger teacher model, suggesting that models can internalize safe and helpful behavior elicited from their own. Warning: This paper contains content that may be offensive or harmful.

06.
arXiv (CS.CL) 2026-06-17

From Observation to Intervention: A Causal Audit of Expert Importance in Mixture-of-Experts Models

Interpretability methods routinely use population-level summary statistics over observed model behaviour to license claims about the effects of targeted interventions on specific computations; in Pearl's terms, they treat rung-1 associational evidence as if it supported rung-2 interventional conclusions, a move whose validity is rarely tested. We examine one concrete instance: the use of routing statistics in Mixture-of-Experts (MoE) pruning, where utilization rates, activation norms, and routing weight distributions are treated as predictors of which experts can be removed without functional cost. A token-level interventional audit across three high-redundancy MoE architectures (OLMoE-1B-7B-0924, Qwen1.5-MoE-A2.7B, DeepSeek-V2-Lite) finds no observational metric predicts causal expert importance in any model: across all 60 metric-layer combinations effect sizes stay below Cohen's $d = 0.23$, and no metric is reliably positive under our corrected, dual-test criterion. A per-token routing weight control, run with identical $n$, rules out insufficient power, recovering a signal whose CI excludes zero at OLMoE's final MoE layer ($d = +0.231$, 95\% CI $[+0.09, +0.37]$, $p = 0.0013$). Existing pruning methods succeed in this regime not by identifying dispensable experts but because early-layer redundancy renders most selection criteria interchangeable. Our results provide an explicit counterexample to the common inferential step from population-level observational summaries to token-level interventional claims about expert importance, and illustrate how interventional audits can calibrate the evidential standards for interpretability claims.

07.
medRxiv (Medicine) 2026-06-22

GCH1 p.Ser80Asn Confers Risk for Parkinson's Disease in East Asian Populations

Introduction: GCH1 has been implicated in Parkinson's disease (PD), but its risks variants and associations are not well defined. Objectives: To investigate the clinical relevance and PD risk associated with the GCH1 p.Ser80Asn variant. Methods: We first identified a segregating GCH1 p.Ser80Asn variant in a Malaysian Chinese PD family via whole genome sequencing (WGS). We assessed its risk association using multi-ancestry WGS data from the Global Parkinson's Genetics Program (GP2) (n=22,372PD vs n=8,826Controls) and meta-analysis of East Asian (EAS) cohorts (n=4,712PD vs 38,733Controls). Clinico-demographic details of affected variant carriers were collated. Results: The GCH1 p.Ser80Asn variant was enriched in GP2 EAS PD populations (n=9/2,757; 0.33%) but not detected in other ancestries. Meta-analysis revealed increased PD risk in EAS populations (odds ratio:5.1; 95%CI:2.3-10.7; p=2.89x10-5). Affected carriers (mean age at onset:56.3+-12.5 years) had additional occurrence of dystonia, while dementia was rare. Conclusions: The GCH1 p.Ser80Asn variant is a rare, EAS-enriched risk variant for PD.

08.
arXiv (quant-ph) 2026-06-16

Against probability: A quantum state is more than a list of probability distributions

arXiv:2601.18872v2 Announce Type: replace Abstract: The state of a quantum system can be represented by listing the outcome probabilities for a tomographically complete set of measurements. Such representations appear throughout physics, for example, in quantum field theory via correlation functions and in quantum foundations within generalized probabilistic frameworks. In this paper, we show a no-go result: To enable useful statements, the probability representation must be topologically robust$\unicode{x2014}$preserving the notion of closeness between states. Yet, a topologically robust probability representation cannot simultaneously retain other essential structure, such as the subsystem structure.

09.
arXiv (CS.CL) 2026-06-11

Every Act Has Its Price: Compressed Moral Composition in Frontier LLMs

Existing LLM moral benchmarks usually ask which isolated moral act, value, or foundation a model prefers. This is useful but incomplete. Realistic judgments often require a model to combine several moral signals within the same option. We introduce **Moral Trolley Arena**, a two-stage blind ELO benchmark for measuring how LLMs compose moral evidence. The single-scene arena first calibrates individual moral acts from a 229-scenario corpus across five Moral Foundations Theory foundations; the composite arena then combines calibrated acts into two-act moral items over a controlled intensity grid and measures the resulting composite preferences. Across ten frontier models, composite judgments are largely predicted by component act strength, but the relation is consistently compressed rather than simply additive. Models also show non-additive intensity anchoring, bounded foundation-specific residuals after component control, and highly convergent composite preference surfaces across providers. These results suggest that moral audits should measure composition rules for moral evidence, not only rankings over isolated acts.

10.
arXiv (CS.CL) 2026-06-18

Possible or Definite? A Benchmark for Evaluating Diagnostic Uncertainty Preservation in Clinical Text

Large language models (LLMs) are increasingly used for clinical text tasks such as summarization and revision. While most studies evaluate the fluency and coherence of LLM-generated text, whether LLMs correctly preserve diagnostic uncertainty remains underexplored. In clinical practice, phrases such as ``possible pneumonia'' communicate the strength of available evidence and directly guide decisions about follow-up testing and treatment. Altering these uncertainty expressions can change the clinical meaning entirely. In this paper, we systematically evaluated this problem in two steps. First, we constructed a benchmark of 1,200 clinical documents with 9,184 uncertainty annotations across five levels. Second, we evaluated three LLMs on this benchmark. Our results show that (1) LLMs preserve the original uncertainty cues poorly, often less than half the time; (2) LLMs struggle with nuanced distinctions between adjacent levels. This work reveals a failure mode not captured by standard evaluation metrics and provides implications for the safe deployment of LLMs in clinical workflows.

11.
arXiv (CS.AI) 2026-06-19

SIMBA: ABidirectional Retrieval Forward Simulation Framework for Modeling FY-4A GIIRS Hyperspectral Infrared Radiances Toward NWP Applications

arXiv:2606.19943v1 Announce Type: cross Abstract: Hyperspectral infrared observations are an important data source for numerical weather prediction (NWP) because they provide rich information on the vertical structure of atmospheric temperature and humidity. However, most existing deep learning methods mainly focus on one-way retrieval from radiances to atmospheric profiles, while the reverse radiance simulation process and the consistency between atmospheric state space and radiance observation space are insufficiently considered. In this study, we propose SIMBA, a unified bidirectional retrieval-forward simulation framework for FY-4A GIIRS hyperspectral infrared radiance modeling toward NWP applications. The framework jointly performs atmospheric profile retrieval and radiance reconstruction, introduces a cycle-consistency constraint to strengthen the coupling between the two processes, and employs a bidirectional Mamba state-space module to capture long-range dependencies along pressure levels. Using collocated FY-4A GIIRS observations and ERA5 reanalysis data, the proposed method is evaluated for temperature retrieval, specific humidity retrieval, long-wave radiance reconstruction, and medium-wave radiance reconstruction. Experimental results show that SIMBA outperforms several representative deep learning baselines across both retrieval and reconstruction tasks, while ablation experiments confirm the contribution of the bidirectional design and cycle-consistency mechanism. These results demonstrate that the proposed framework is effective for joint atmospheric profile retrieval and hyperspectral infrared radiance modeling, and suggest potential for future Jacobian-related analysis and NWP-oriented extensions.

12.
arXiv (CS.CV) 2026-06-25

Generalised Medical Phrase Grounding

Medical phrase grounding (MPG) maps textual descriptions of radiological findings to corresponding image regions. These grounded reports are easier to interpret, especially for non-experts. Existing MPG systems mostly follow the referring expression comprehension (REC) paradigm and return exactly one bounding box per phrase. Real reports often violate this assumption. They contain multi-region findings, non-diagnostic text, and non-groundable phrases, such as negations or descriptions of normal anatomy. Motivated by this, we reformulate the task as generalised medical phrase grounding (GMPG), where each sentence is mapped to zero, one, or multiple scored regions. To realise this formulation, we introduce the first GMPG model: MedGrounder. We adopted a two-stage training regime: pre-training on report sentence–anatomy box alignment datasets and fine-tuning on report sentence–human annotated box datasets. Experiments on PadChest-GR and MS-CXR show that MedGrounder achieves strong zero-shot transfer and outperforms REC-style and grounded report generation baselines on multi-region and non-groundable phrases, while using far fewer human box annotations. Finally, we show that MedGrounder can be composed with existing report generators to produce grounded reports without retraining the generator.

13.
arXiv (CS.CV) 2026-06-16

Beyond Self-Attention: Sub-Quadratic Vision Transformers for Fast Image Captioning

Image captioning is a challenging and significant task that aims to generate coherent and semantically meaningful textual descriptions for given images. To accomplish this task, it requires a deep understanding of visual content along with the ability to express that understanding in natural language. Despite remarkable progress with transformer-based architectures, existing approaches often suffer from limitations, such as a lack of rich local feature representations and the high computational cost of quadratic self-attention. The proposed model focuses on improving computational efficiency by restructuring the vision transformer architecture. In designing this approach, the standard self-attention mechanism in Vision Transformers is replaced with a probabilistic transformer approach based on a Gaussian Mixture Model (GMM), a soft-clustering technique. Instead of computing pairwise attention among all image patches, the model groups similar patches into a fixed number of clusters using an Expectation-Maximization (EM) algorithm. This clustering-based mechanism reduces the computational complexity from quadratic O(n^2) to linear O(nK), where K

14.
arXiv (CS.AI) 2026-06-12

Neuro-Symbolic Agents for Regulated Process Automation: Challenges and Research Agenda

arXiv:2606.13405v1 Announce Type: new Abstract: LLM-based agents are entering regulated industries where they automate judgment intensive quality management processes. We argue that symbolic structures already embedded in these domains, including regulations, typed process models, and compliance constraints, should be treated not merely as external monitoring mechanisms but as core architectural components that shape the agent's decision-making and behavior. We propose compliance-by-construction as a complementary paradigm to guardrail-based monitoring: a structural foundation that prevents control-flow violations, while guardrails remain essential for catching semantic errors. We identify a structured set of neuro-symbolic research challenges on foundational and capability level and show that addressing them jointly enables compliance-by-construction. We call on the neuro-symbolic community to engage with regulated process automation as a high impact research domain.

15.
arXiv (CS.CL) 2026-06-25

Adaptive Oscillatory Inductive Bias for Modeling Sharp Prosodic Dynamics in Diffusion-Based TTS

Diffusion-based text-to-speech (TTS) models have achieved significant improvements in speech quality. However, modeling sharp prosodic transitions and rapid pitch variations in expressive speech remains challenging. Existing diffusion-based TTS decoders commonly utilize periodic nonlinearities such as Snake activation function to capture harmonic structures, but this activation funcation provides limited adaptability when modeling abrupt amplitude and frequency variations. In this paper, we investigate the role of oscillatory inductive bias in diffusion-based TTS decoders and introduce an adaptive oscillatory nonlinearity that enables controllable periodic modulation while maintaining signal stability through a linear bypass component. We refer the resulting TTS system as OscillaTTS. Experiments on the LJSpeech and Emotional Speech Dataset show consistent improvements across objective and subjective evaluations, indicating improved modeling of expressive prosodic dynamics.

16.
arXiv (CS.AI) 2026-06-16

AL-GNN: Privacy-Preserving and Replay-Free Continual Graph Learning via Analytic Learning

arXiv:2512.18295v2 Announce Type: replace-cross Abstract: Continual graph learning (CGL) aims to enable graph neural networks to incrementally learn from a stream of graph structured data without forgetting previously acquired knowledge. Existing methods particularly those based on experience replay typically store and revisit past graph data to mitigate catastrophic forgetting. However, these approaches pose significant limitations, including privacy concerns, inefficiency. In this work, we propose AL GNN, a novel framework for continual graph learning that eliminates the need for backpropagation and replay buffers. Instead, AL GNN leverages principles from analytic learning theory to formulate learning as a recursive least squares optimization process. It maintains and updates model knowledge analytically through closed form classifier updates and a regularized feature autocorrelation matrix. This design enables efficient one pass training for each task, and inherently preserves data privacy by avoiding historical sample storage. Extensive experiments on multiple dynamic graph classification benchmarks demonstrate that AL GNN achieves competitive or superior performance compared to existing methods. For instance, it improves average performance by 10% on CoraFull and reduces forgetting by over 30% on Reddit, while also reducing training time by nearly 50% due to its backpropagation free design.

17.
arXiv (CS.CV) 2026-06-25

Streaming-dLLM: Accelerating Diffusion LLMs via Suffix Pruning and Dynamic Decoding

Diffusion Large Language Models (dLLMs) offer a compelling paradigm for natural language generation, leveraging parallel decoding and bidirectional attention to achieve superior global coherence compared to autoregressive models. While recent works have accelerated inference via KV cache reuse or heuristic decoding, they overlook the intrinsic inefficiencies within the block-wise diffusion process. Specifically, they suffer from spatial redundancy by modeling informative-sparse suffix regions uniformly and temporal inefficiency by applying fixed denoising schedules across all the decoding process. To address this, we propose Streaming-dLLM, a training-free framework that streamlines inference across both spatial and temporal dimensions. Spatially, we introduce attenuation guided suffix modeling to approximate the full context by pruning redundant mask tokens. Temporally, we employ a dynamic confidence aware strategy with an early exit mechanism, allowing the model to skip unnecessary iterations for converged tokens. Extensive experiments show that Streaming-dLLM achieves up to 68.2X speedup while maintaining generation quality, highlighting its effectiveness in diffusion decoding. The code is available at https://github.com/xiaoshideta/Streaming-dLLM.

18.
arXiv (CS.CL) 2026-06-16

A Mechanistic Understanding of Pronoun Fidelity in LLMs

Faithful and robust pronoun use is important for fair and coherent generations, yet large language models largely fail when multiple referents use different pronouns. To study the interplay of reasoning, repetition, and bias in this task, prior work relies exclusively on behavioural approaches, which may not reflect a model's internal workings. Therefore, we provide a mechanistic, model-internal perspective on pronoun fidelity, testing whether three mechanisms – group entity binding (G), recency bias (R), and stereotypical bias (S) – are causally implemented across several SOTA language models. Using Boundless Distributed Alignment Search, we find all three coexist as causal subspaces distributed across network depth. No single mechanism fully explains model behaviour, but a combination of the three consistently accounts for 91-99.5%. An attention head analysis further reveals two competing copying routes; group binding and stereotype share a localized concept-level route that retrieves a bound occupation-pronoun unit, while recency uses a distributed token-level route that repeats surface forms. In sum, pronoun fidelity arises from competition between simultaneously active causal subspaces.

19.
arXiv (CS.LG) 2026-06-18

A Streaming Sparse Cholesky Method for Derivative-Informed Gaussian Process Surrogates Within Digital Twin Applications

arXiv:2511.00366v2 Announce Type: replace-cross Abstract: Digital twins are developed to model the behavior of a specific physical asset (or twin), and they can consist of high-fidelity physics-based models or surrogates. A highly accurate surrogate is often preferred over multi-physics models as they enable forecasting the physical twin future state in real-time. To adapt to a specific physical twin, the digital twin model must be updated using in-service data from that physical twin. In this paper, we combine and extend several previous surrogate-related advancements with the goal of demonstrating an end-to-end digital twin (DT) solution for predicting performance of an aircraft structure (the physical asset). To this end, we extend Gaussian process (GP) models to include derivative data, for improved accuracy, with dynamic updating to ingest physical twin data during service. Including derivative data, however, comes at a prohibitive cost of increased covariance matrix dimension. We circumvent this issue through our modified dynamic sparse Cholesky linear system solver. Numerical experiments demonstrate that the prediction accuracy of the derivative-enhanced sparse Cholesky GP method produces improved models upon dynamic data additions. Lastly, we demonstrate the developed algorithm within a DT framework to model fatigue crack growth in an aerospace vehicle, thereby exhibiting through our assembled engineered system how digital twin technologies can be combined in practice.

20.
arXiv (CS.AI) 2026-06-17

Prefill/Decode-Aware Evaluation of LLM Inference on Emerging AI Accelerators

arXiv:2606.17104v1 Announce Type: cross Abstract: As large language models (LLMs) are increasingly deployed in latency- and cost-sensitive settings, inference efficiency has become a central systems challenge. While GPUs dominate current deployments, a growing number of AI accelerators claim advantages for LLM inference, yet it remains unclear under which conditions such accelerators outperform GPUs in practice. Recent inference systems decompose execution into Prefill and Decode phases, which exhibit distinct computational characteristics and latency metrics, commonly captured by time to first token (TTFT) and time per output token (TPOT). This paper presents a phase-aware evaluation of LLM inference performance across GPUs and emerging AI accelerators using a common model, Llama2-7B. By separately measuring Prefill and Decode performance, we reveal that accelerator advantages differ by phase and metric. Our results show that GPUs consistently excel in the compute-intensive Prefill phase, while GroqRack achieves significantly lower TPOT during Decode (batching not currently supported). However, GPUs regain an advantage in Decode throughput as batch size increases. These findings demonstrate that each platform exhibits distinct phase-dependent strengths. We further analyze heterogeneous Prefill/Decode disaggregation across different accelerator platforms, identifying performance gains and the workload and network conditions under which such gains are realized.

21.
arXiv (CS.CV) 2026-06-16

Disagreement-Based Cross-Model Routing for Implicit Video Question Answering

We study multiple-choice video question answering on the ImplicitQA benchmark, where the correct answer is never explicitly shown but must be inferred from off-screen events, line-of-sight cues, causal structure, and cross-shot spatial layout. On this benchmark a single frontier video LLM already operates near its accuracy ceiling, and we observe that conventional self-consistency strategies – majority voting across repeated samples of the same model – can hurt rather than help, because the model's errors on hard questions are correlated. We propose disagreement-based cross-model routing, a pure inference-time procedure that requires no labels and no training. We triple-sample a native-video model (Gemini 3.1 Pro Preview) at temperature zero, exploit the genuine sample-to-sample variance of its video-processing pipeline to identify the roughly 20% subset of questions where the three samples disagree, and route only that subset to a second model from a different family (Claude Opus 4.8) that consumes uniformly sampled frames with adaptive thinking. On the 1001-question validation set with public ground truth – our main evaluation – the method improves AvgAcc by +1.43 over the best single sample of the primary model, with per-category gains concentrated on Motion & Trajectory (+5.49), Inferred Counting (+3.45), and Vertical Spatial Reasoning (+1.82) – the categories most dependent on cross-shot reference resolution. The same pipeline applied to the held-out 172-question CVPR 2026 ImplicitQA challenge test set achieves 82.03 AvgAcc / 79.71 MacroAvgAcc (+1.81 over the best single sample of the primary model), confirming the validation result on an independent split.

22.
arXiv (CS.LG) 2026-06-25

A Geometry-Aware Efficient Algorithm for Compositional Entropic Risk Minimization

arXiv:2602.02877v2 Announce Type: replace Abstract: This paper studies optimization for a family of problems termed $compositional entropic risk minimization$, in which each data's loss is formulated as a Log-Expectation-Exponential (Log-E-Exp) function. The Log-E-Exp formulation serves as an abstraction of the Log-Sum-Exponential (LogSumExp) function when the explicit summation inside the logarithm is taken over a gigantic number of items and is therefore expensive to evaluate. While entropic risk objectives of this form arise in many machine learning problems, existing optimization algorithms suffer from several fundamental limitations including non-convergence, numerical instability, and slow convergence rates. To address these limitations, we propose a geometry-aware stochastic algorithm, termed $SCENT$, for the dual formulation of entropic risk minimization cast as a min–min optimization problem. The key to our design is a $stochastic proximal mirror descent (SPMD)$ update for the dual variable, equipped with a Bregman divergence induced by a negative exponential function that faithfully captures the geometry of the objective. Our main contributions are threefold: (i) we establish an $O(1/\sqrt{T})$ convergence rate of the proposed SCENT algorithm for convex problems; (ii) we theoretically characterize the advantages of SPMD over standard SGD update for optimizing the dual variable; and (iii) we demonstrate the empirical effectiveness of SCENT on extreme classification, partial AUC maximization, contrastive learning and distributionally robust optimization, where it consistently outperforms existing baselines. Code is available at https://github.com/Optimization-AI/SCENT.

23.
arXiv (CS.CL) 2026-06-15

Knowing When to Quit: A Principled Framework for Dynamic Abstention in LLM Reasoning

LLMs utilizing chain-of-thought reasoning often waste substantial compute by producing long, incorrect responses. Abstention can mitigate this by withholding outputs unlikely to be correct. While most abstention methods decide to withhold outputs before or after generation, dynamic mid-generation abstention considers early termination of unpromising reasoning traces at each token position. Prior work has explored empirical variants of this idea, but principled guidance for the abstention rule remains lacking. We present a formal analysis of dynamic abstention for LLMs, modeling abstention as an explicit action within a regularized reinforcement learning framework. An abstention reward parameter controls the trade-off between compute and information. We show that abstaining when the value function falls below this reward strictly outperforms natural baselines under general conditions. We further derive a principled and efficient method to approximate the value function. Empirical results on mathematical reasoning and toxicity avoidance tasks support our theory and demonstrate improved selective accuracy over existing methods.

24.
arXiv (CS.CV) 2026-06-17

MLLMs Get It Right, Then Get It Wrong: Tracing and Correcting Late-Layer Textual Bias

When vision contradicts text, multimodal large language models (MLLMs) consistently favor text, even when images provide clear evidence otherwise. This bias poses risks for applications requiring visual grounding, yet its cause remains unclear. In this paper, we uncover a surprising finding: models often get it right initially, forming correct vision-based predictions in their intermediate layers, before changing their minds and favoring text in the final output. We call this "late-layer textual override". The visual information is encoded, it simply does not survive to the output. More intriguingly, we find that how predictions change reveals whether they're correct: 85% of failures shift toward text, while 89% of successes shift toward vision. This directional signature enables a simple but powerful intervention: when we detect a confident visual prediction being suppressed, we restore it. We propose CALRD (Conflict-Aware Layer Reference Decoding), a training-free method that recovers overridden predictions at inference time. Experiments across five MLLMs of varying architectures demonstrate up to 9.4% absolute improvements on conflict benchmarks while largely preserving standard performance, without training or external knowledge. It recovers what the model already knew but failed to preserve.

25.
arXiv (CS.AI) 2026-06-24

Does Mixture-of-Experts Actually Help Inference on Consumer and Edge Hardware? An Empirical Study

arXiv:2606.21428v2 Announce Type: replace-cross Abstract: Mixture-of-Experts (MoE) language models are often described as ideal for resource-constrained inference. Each token activates only a small subset of experts, so the per-token compute cost, in floating-point operations (FLOPs), resembles that of a much smaller dense model. Whether that FLOP advantage survives in practice is far less clear. We ask whether MoE models actually run faster and cheaper than comparable dense models on consumer-grade and edge hardware. We benchmark OLMoE-1B-7B (1.3 B active of 6.9 B total) against three dense baselines on an Apple M2 Pro and an NVIDIA Jetson Orin Nano 8 GB through \texttt{llama.cpp}, measuring throughput, memory, and on-device energy. The answer is device-dependent: OLMoE's active-parameter advantage is only partly realised on the laptop (~10% behind the same-active Llama-3.2-1B) and erodes on the edge device (~31% behind, at 2.1$\times$ the energy per token, with peak memory at the 8 GB ceiling). Patching \texttt{llama.cpp} to time the decode graph node-by-node shows routing accounts for under 9% of MoE-block compute on the cleaner edge backend, so the gap reflects total-parameter memory footprint, expert dispatch, and KV-cache pressure rather than routing. The implication is that on bandwidth-bound edge hardware, inference cost tracks total parameters, not active ones, and sparse activation does not buy back what the device is constrained on. These findings are bounded to one MoE model at this parameter scale and two devices, and we release the full measurement harness and per-run data.