NeurIPS 2026 Papers

Most Vision-Language Models (VLMs) are built by extending pretrained Large Language Models (LLMs) with visual modules and multimodal alignment. However, this multimodal scaling often degrades the language-side reasoning ability originally encoded in the base LLM. While the base LLM retains usable reasoning ability after scaling, the aligned VLM itself cannot reliably access this ability. Therefore, recovering degraded reasoning capability in VLMs may be more effective when using the base LLM as the source rather than relying on the VLM alone. Motivated by this, we propose **LIFT** (**L**anguage-side reason**i**ng **F**acilitation and **T**ransfer), a lightweight vector-intervention method that transfers reasoning capability from the base LLM to the VLM without retraining the backbone. LIFT defines **Reasoning Vectors** as answer-token hidden-state differences between a Reasoner path with an explicit reasoning trace and a Solver path without it, and injects these vectors into language-side activations of the target VLM. LIFT further supports learnable vector adaptation while keeping the VLM backbone frozen. We evaluate LIFT on two VLMs across six reasoning benchmarks, comparing Reasoning Vectors extracted from the base LLM and from the aligned VLM under matched protocols. Results show that LLM-derived vectors consistently outperform VLM-derived vectors, confirming that the base LLM is a more effective source for recovering reasoning. LIFT partially recovers degraded reasoning through lightweight language-side interventions. Further analyses show that Reasoning Vectors influence intermediate reasoning behavior rather than merely altering final answers.

Recently, video generation models have achieved remarkable progress in global visual fidelity. However, when synthesizing complex dynamic content, they frequently produce **sparse subtle anomalies**. Existing video reward models primarily emphasize overall visual quality and text alignment; yet as generation quality improves, these coarse-grained defects become less frequent, while the sparse structural, temporal, and physical anomalies become an increasingly important bottleneck. Therefore, in this paper, we propose Concentrate and Concentrate (CaC), a coarse-to-fine anomaly reward model based on Vision-Language Models. During inference, it first conducts a global temporal scan to anchor anomalous time windows, then performs fine-grained spatial grounding within the localized interval, and finally derives robust judgments via structured spatiotemporal Chain-of-Thought reasoning. To equip the model with these capabilities, we construct the first large-scale generated video anomaly dataset with per-frame bounding-box annotations, temporal anomaly windows, and fine-grained attribution labels. Building on this dataset, we design a three-stage progressive training paradigm. The model initially learns spatial and temporal anchoring through single- and multi-frame supervised fine-tuning, and then is optimized by a reinforcement learning strategy based on two-turn Group Relative Policy Optimization (GRPO). Beyond conventional accuracy rewards, we introduce Temporal and Spatial IoU rewards to supervise the intermediate localization process, effectively guiding the model toward more grounded and interpretable spatiotemporal reasoning. Extensive experiments demonstrate that CaC can stably ``concentrate'' on subtle anomalies, achieving a 25.7\% accuracy improvement on fine-grained anomaly benchmarks and, when used as a reward signal, CaC reduces generated-video anomalies by 11.7\% while improving overall video quality.

Visual reprogramming (VR) efficiently repurposes pre-trained vision-language models for new image classification tasks by adding trainable patterns to inputs without modifying the backbone. Nevertheless, existing VR methods heavily depend on labeled data and thus suffer substantial performance degradation under limited supervision (i.e., scarce labeled images). We introduce semi-supervised learning (SSL) into VR to exploit abundant unlabeled images and enhance data efficiency. However, directly applying generic SSL techniques often amplifies biases in unlabeled posteriors and destabilizes VR training. To address these challenges, we propose **CURE** (**C**onsistency and **U**nlabeled **R**ecalibration with Confidence-Margin **E**nhancement), a semi-supervised VR framework comprising two key components: *(i)* Recalibreted Gaussian Confidence-Margin Soft Weighting to dynamically adjust sample importance; *(ii)* Dual-attribute Consistency Regularization to ensure consistency across attribute prompts. Extensive experiments show that CURE consistently outperforms both existing VR methods and direct SSL extensions, improving average classification accuracy by **5.06** and **6.67** percentage points across twelve widely-used benchmarks for limited supervision tasks. Our code is available at https://anonymous.4open.science/r/CURE-93E2/.

Despite remarkable advances, today's AI systems remain narrow in scope, falling short of the flexible, adaptive, and multisensory intelligence that characterizes human capabilities. This gap has fueled longstanding debates about whether AI might one day achieve human-like generality or even consciousness, and whether theories of consciousness can inspire new architectures for AI. This paper presents an early blueprint for implementing a general AI system, CTM-AI, combining the Conscious Turing Machine (CTM), a formal machine model of consciousness, with today's foundation models. CTM-AI contains an enormous number of powerful processors ranging from specialized experts (e.g., vision-language models and APIs) to unspecialized general-purpose learners poised to develop their own expertise. Crucially, for whatever problem must be dealt with, information from many processors is selected, integrated, and exchanged appropriately to solve the task.CTM-AI achieves state-of-the-art accuracy on MUStARD (72.28) and UR-FUNNY (72.13), outperforming multimodal and multi-agent frameworks. On tool-using and agentic tasks, CTM-AI achieves 10+ points of improvement on StableToolBench and WebArena-Lite. Overall, CTM-AI offers a principled, testable blueprint for general AI inspired by a model of consciousness.

Indoor context deciphered by ambient sensing underpins many applications, from assisted living to smart buildings, yet remains weakly connected to the language-aligned paradigm of modern multimodal systems. Through billions of IoT devices, ambient WiFi Channel State Information (CSI) captures the indoor context of human activity and environmental dynamics passively via ubiquitous, always-on infrastructure. However, unlike vision or audio, CSI lacks naturally paired text like captions or transcripts, limiting its integration with language-driven AI systems. We present CSI-TextBench, a large-scale dataset and benchmark for aligning CSI data with natural language, unifying 3 sensing benchmarks into 285,790 samples across 7 tasks and pairing them with 2,291 hierarchically structured, label-derived descriptions via many-to-many assignment to construct 2.0 million CSI–text training tuples. Benchmarking 5 alignment methods from 4 paradigm families, we show that contrastive CSI--text alignment matches supervised classifiers without class labels, adapts to unseen classes with few samples, and responds to open-vocabulary natural-language prompts. A single unified model supports compositional scene understanding, answering joint queries over activity, proximity, and identity, achieving 86.9\% top-1 accuracy on 140 composite scenes, outperforming task-specific classifier ensembles. CSI-TextBench enables ubiquitous ambient sensing as a language-indexed perception layer, enabling indoor contextual awareness in language-driven AI systems.

While LLM-based coding agents have advanced to repository-scale engineering, existing benchmarks predominantly focus on software maintenance, overlooking the foundational capability of system construction. We introduce CSBench, a benchmark evaluating project-level Spec-to-Code implementation across core computer science domains (e.g., OS, compilers, databases). CSBench features 100 expert-curated tasks from top-tier university assignments within containerized environments. Evaluating 12 state-of-the-art LLMs reveals a stark performance disparity: even top-tier models like GPT-5.2 see success rates plummet from over 90% in algorithmic tasks to below 35% in system-level domains. Our analysis identifies a fundamental deficiency in systemic reasoning and construction knowledge as the primary bottleneck, where agents either terminate prematurely due to overconfidence or fail to translate execution feedback into valid architectural fixes, highlighting CSBench as a necessary tool to bridge this gap and advance autonomous agents toward deep computer science mastery.

Code agents must both reason over long-horizon repository state and obey strict tool-use protocols. In paired Instruct/Thinking checkpoints, these capabilities are complementary but misaligned. The Instruct model is concise and tool-disciplined, whereas the Thinking model offers stronger planning and recovery behavior but often over-deliberates and degrades agent performance. We present CRANE (Constrained Reasoning Injection for Code Agents via Nullspace Editing), a training-free parameter-editing method that treats the Thinking–Instruct delta as a directional pool of candidate reasoning edits for the Instruct backbone. CRANE combines magnitude thresholding to denoise the delta, a Conservative Taylor Gate to retain edits that are jointly beneficial for reasoning transfer and tool-use preservation, and Graduated Sigmoidal Projection to suppress format-critical update directions. By merging paired Instruct and Thinking checkpoints, CRANE delivers strong gains over either individual model while preserving Instruct-level efficiency: on Roo-Eval it achieves pass@1 of 66.2% (+19.5%) for Qwen3-30B-A3B and 81.5% (+8.7%) for Qwen3-Next-80B-A3B; on SWE-bench-Verified it resolves up to 14 additional instances at both scales (122/500 and 180/500); and on Terminal-Bench v2 it improves pass@1/pass@5 by up to 2.3%/7.8%, reaching 7.6%/17.9% and 14.8%/30.3%, respectively, consistently outperforming alternative merging strategies across all three benchmarks.

Open-loop imitation learning has advanced modern autonomous driving policy architectures, but closed-loop deployment remains vulnerable to policy-induced distribution shift. Existing post-training paradigms exhibit fundamental trade-offs: closed-loop RL fine-tuning provides grounded feedback from executed actions but is constrained by the sparsity of informative events, whereas counterfactual fine-tuning provides dense supervision over candidate futures but inherits bias from imperfect future estimates. We introduce Counterfactual-to-Interactive Reinforcement Fine-Tuning (CRAFT), an on-policy framework that formulates closed-loop post-training as proxy-residual optimization. CRAFT uses group-normalized counterfactual advantages as a dense proxy for real closed-loop advantages and aligns this proxy with the closed-loop world through grounded residual correction from interaction-critical events. To stabilize adaptation, CRAFT regularizes the online policy toward an EMA teacher via asymmetric KL self-distillation. Theoretically, CRAFT decomposes the real closed-loop policy gradient into proxy and residual terms under the same visited-state distribution, reducing residual variance with an aligned proxy while mitigating proxy bias through grounded residual approximation. Empirically, CRAFT achieves the strongest closed-loop gains on Bench2Drive across hierarchical planning, vision-language-action, and vocabulary-scoring architectures. Ablations, scaling behavior, stability analyses, and transfer results further validate the complementary roles of dense counterfactual proxy and grounded residual correction.

As vision language models are increasingly deployed in clinical diagnosis, understanding how they internally resolve competing visual and textual signals becomes a safety imperative. Existing mechanistic analyses remain confined to unimodal text and offer no explanation for why a single misleading sentence can override a correct image based diagnosis, or why a model commits to a confident answer despite insufficient visual evidence. We find that these two safety risks, arbitration failure where textual context overrides visual grounding and brake failure where the model commits without adequate evidence, are mediated by spatially disjoint attention head populations: arbitration heads form a mid-to-deep wideband reflecting cross-layer evidence competition, while brake heads concentrate in a narrow middle-to-late layer band that regulates evidence sufficiency and abstention behavior. To ground these observations in causal circuitry, we introduce CRAFT, which localizes each failure mode to a minimal causal head set via dual criteria and verifies necessity and sufficiency through temporal probes and Tuned Lens trajectory analysis. Excising arbitration heads sharply reduces conflict following with negligible degradation on clean inputs, while excising brake heads restores appropriate abstention under degraded visual evidence. The two interventions target spatially disjoint head sets and produce distinct corrective effects, underscoring the mechanistic separability of the failure modes. Experiments across multiple medical VQA benchmarks and VLM architectures validate both the localization and interventions, demonstrating that the identified heads causally drive each failure mode and that targeted modulation generalises without retraining. The code is available at \url{https://anonymous.4open.science/r/CRAFT-B3CA}.

Automatic neuron reconstruction from light microscopy images is a central problem in computational neuroanatomy. While recent methods have achieved encouraging results on local image blocks, it remains unclear whether such progress translates to reconstruction that is both structurally accurate and scalable to the whole-brain scale. We present CORAL, the first benchmark for structure-aware evaluation of automatic neuron reconstruction from light microscopy images at both local and whole-brain scales. Built on a high-quality whole-brain fMOST dataset with carefully curated annotations, CORAL establishes two progressive tasks: block-level reconstruction, which evaluates reconstruction methods under limited spatial context, and brain-wide reconstruction, which assesses complete neuron reconstruction at the whole-brain scale. To account for topological correctness beyond geometric distance similarity, we introduce a structure-aware metric based on fiber prediction. To further achieve complete neuron reconstruction across the entire brain, we develop a brain-wide neuron tracing framework that extends arbitrary local reconstruction methods to the whole-brain scale through an iterative local-to-global process. Using this benchmark, we provide the first structure-aware comparison of mainstream methods for local neuron reconstruction and further evaluate their performance in brain-wide reconstruction. Our results underscore the importance of structure-aware evaluation and the need for more robust methods for complete neuron reconstruction. All datasets and code are available at https://github.com/yang-ze-kang/CORAL.

Label-free proxy scores are often used as if they measured downstream discriminative difficulty. A score may rank samples similarly to a target, predict that target on held-out data, or improve a downstream decision; these are distinct evidence levels. We present Cross-Objective Hardness Evaluation (COHE), a protocol that audits each proxy--target--protocol tuple through dependence, held-out-prediction, and operational-transfer gates. COHE maps standard, inspectable statistics to bounded claim tiers; the contribution is evidence-to-claim calibration, not a new sampler, metric, or model. A DINOv2 isolation stress test illustrates the boundary: representation isolation captures a first-learning dynamics relation (rho = 0.4363, R^2 = 0.2126), but using it for direct 10% subset selection fails against Random (Delta = -15.41, 95% CI [-15.87, -14.95]). COHE therefore preserves this target-specific relation without rebranding it as actionability. Across CIFAR-10/100, ImageNet-10, and ImageNet-1K validation checks, eight primary generative proxy families do not support CE/margin surrogate claims: they show weak alignment, near-zero held-out CE prediction, and unreliable transfer. Controls show stable discriminative target rankings, expected shuffled-proxy null behavior, and a task-aligned ImageNet target-hard retrieval/triage tuple that passes all gates. The anonymized supplementary package provides lightweight gate-wise audit code, toy scalar CSV examples, cached aggregate summaries, selected derived scalar arrays for tabular gate verification, supported paper-style verification scripts, and provenance notes without redistributing raw datasets, images, weights, checkpoints, or full scoring/retraining pipelines.

We formalize and enable the task of open tree decomposition, which segments an image into hierarchical trees of visual components with unconstrained granularity and flexibility. Specifically, we provide the foundation benchmark for this new paradigm with the following three key contributions. First, we overcome the prohibitively high cognitive and physical bottlenecks of manual annotation by developing a fully automated generation pipeline that synergizes the semantic reasoning of Large Vision-Language Models (LVLMs) with the precise geometric grounding of SAM 3. Second, leveraging this pipeline, we construct COCOTree, a massive-scale benchmark featuring over 21K images and 1.8M structural nodes. By embracing an open-vocabulary space of over 3.5K unique labels, it successfully captures the long-tail distribution of complex physical assemblies. Notably, rigorous human evaluation confirms our generated annotations demonstrate strong alignment with human structural judgment. Third, we establish a standardized evaluation protocol by proposing the Open Tree Quality (OTQ) metric, which jointly assesses mask precision, label accuracy, and structural consistency. We release our dataset and benchmark code at https://github.com/melonkick3090/COCOTree.

Modern 3D visual learning relies on observations sampled from metric 3D assets, yet existing scans, meshes, point clouds, simulations, and reconstructions do not directly provide a sparse, comparable, and geometry-consistent panoramic training interface. Dense trajectories duplicate nearby views, source-specific rendering policies yield heterogeneous annotations, and sparse heuristics may miss important regions or introduce depth-inconsistent observations. *We study how to convert 3D assets into sparse panoramic RGB-D-pose data that preserves complete scene coverage with low redundancy and auditable provenance.* We propose **COVER** (**C**overage-**O**riented **V**iewpoint curation with **E**RP **R**ange-depth warping), a training-free ERP viewpoint curator that projects geometry observed from selected views into candidate ERP probes, scores incremental coverage, and penalizes depth conflicts. Under bounded proxy error, its greedy coverage proxy preserves the standard coverage-style approximation behavior up to an additive error term. Using COVER, we build **CM-EVS** (**C**overage-curated **M**etric **E**RP **V**iew **S**et), a panoramic RGB-D-pose dataset with 36,373 curated ERP frames from 1,275 indoor scenes across Blender indoor, HM3D, and ScanNet++, complemented by outdoor panoramas from TartanGround and OB3D re-encoded into the same schema. Each frame provides full-sphere RGB, metric range depth along ERP rays, calibrated world-to-camera pose, and provenance logs. With a median of only 25 frames per indoor scene, CM-EVS covers all 13 unified room types while maintaining compact scene-level coverage. Experiments show that COVER improves the coverage–conflict trade-off, making CM-EVS a sparse, compact, and auditable RGB-D-pose resource for geometry-consistent panoramic 3D learning.

Style transfer aims to render target content in the style of a reference image, but existing methods often suffer from content leakage, where objects, layouts, or semantics from the style reference appear in the generated output. Although prior data-driven and training-free methods can reduce leakage, they often face a leakage--degradation dilemma: stronger content suppression may weaken style fidelity, while richer style preservation may reintroduce unwanted reference content. We identify this dilemma across the full style-transfer pipeline, including feature separation, feature-space grounding, and diffusion generation. To address these issues, we propose CLeaR, a training-free framework for content-leakage-resistant style transfer. CLeaR first uses Orthogonal Subspace Projection to define content-reduced style targets in each vision foundation model (VFM) feature space. It then performs Ensemble Inversion, which optimizes a shared pixel-space style anchor satisfying style constraints across multiple VFMs. Finally, Energy-Guided Calibration maintains style alignment during diffusion sampling by steering the denoising trajectory toward the ensemble-defined style manifold. We further provide a theoretical analysis showing that the style-anchor estimation error decreases with the number of VFMs. Experiments on StyleBench demonstrate that CLeaR improves style alignment, reduces content leakage, and achieves better LLM-as-Judge evaluation compared with existing methods.

Clinical EEG interpretation requires reasoning over full EEG sessions and integrating signal patterns with clinical context. Existing EEG foundation models are largely designed for short-window decoding and do not incorporate clinical context. We introduce CLEF, a clinically grounded long-context EEG foundation model. CLEF represents EEG sessions as 3D multitaper spectrogram tokens, enabling tractable Transformer modeling at session scale, and aligns embeddings with neurologist reports and structured EHR data through contrastive objectives. We evaluate CLEF on a new 234-task benchmark spanning disease phenotypes, medication exposures, and EEG findings, with more than 260k EEG sessions from over 108k patients. CLEF outperforms prior EEG foundation models on 229 of 234 tasks, improving mean AUROC from 0.65 to 0.74. Reconstruction-only pretraining surpasses prior EEG foundation models, while report and EHR alignment yields further gains. Held-out concept and external-cohort experiments suggest that these representations transfer beyond observed alignment targets. These results support session-scale, clinically grounded representation learning as a promising foundation-model paradigm for clinical EEG.

We introduce CIPHERGRID, a rule-inference and path-finding benchmark. In each task, models are given example grid-world images paired with encoded descriptions and encoded solutions. From this evidence, models then receive a new encoded grid-world to solve. To answer correctly, models must infer which symbols denote tiles, actions, row boundaries, and game mechanics. They must then decode the new world, and produce an encoded path from the start cell to the target cell. The benchmark is designed so that each component is individually solvable: rule inference is calibrated against human performance, and path-finding is solver-verifiable via exact shortest-path search. Although the underlying skills are individually solvable and verifiable, CIPHERGRID tests their interaction: whether models can infer an unfamiliar symbolic system, and carry it through rule application, and planning. CIPHERGRID remains difficult for current frontier models, which only achieve between 18.74\% and 41.89\% accuracy. CIPHERGRID offers a controlled test of whether models can coordinate multimodal grounding, rule inference, and sequential planning under unknown symbolic relationships.

Despite growing interest in on-device LLM deployment, large-scale Mixture-of-Experts (MoE) models remain impractical on resource-constrained devices, as sparse expert activation still requires all expert weights to be memory-resident. To address this, we identify three key observations for on-device MoE inference: (a) cache hit rates are fundamentally bounded even under Belady’s optimal policy; (b) these bounds are insufficient for low-latency on-device inference; and (c) prior cache-aware routing based on binary cache presence can induce routing instability. Based on these insights, we propose CECAR (Cache & Expert Co-Aware Routing), a MoE inference system that jointly considers expert routing and cache management. Unlike prior cache-aware routing methods, CECAR uses an ML-based cache policy to predict the expected reuse distance of each expert and prioritizes routing and execution toward experts with smaller predicted reuse distances. On the Qwen3-30B-A3B model, CECAR achieves 12.5 tokens/s on a consumer-grade GPU with only 16 GB of VRAM, where non-resident experts are fetched from SSD, delivering a 4.5× speedup over conventional MoE inference and exceeding the decode speed of baselines that keep the entire model fully resident in DRAM.

Set-level visual difference discovery is increasingly important for dataset auditing and for understanding model behavior under distribution shift, yet manually inspecting thousands of images is impractical. This motivates the emerging task of *set difference captioning*, i.e., describing concepts that are more often true for one image set than another. Different from existing approaches which heavily rely on large language models (LLMs), incurring substantial computational cost in terms of time and tokens, we introduce **CCDiff**, a lightweight, statistically grounded, and training-free framework for set difference captioning that significantly reduces LLM dependence during core difference discovery, achieving over **$2\times$** speedup and reducing token usage to zero. CCDiff operates in three stages: constructing a shared pool of candidate concepts from domain vocabularies or lightweight captioning models, filtering candidates to obtain set-specific concept pools, and performing inverse canonical correlation analysis (CCA) across sets to identify low-correlation directions that correspond to set differences. Candidate concepts are then ranked using a CCDiff score that balances inverse correlation with within-set representativeness. We evaluate CCDiff on VisDiffBench and further validate it on additional domains including MetaShift and MIMIC-CXR. Across benchmarks and real-world settings, CCDiff matches or exceeds prior LLM-based approaches while being substantially more efficient and scalable.

Vision-Language Models (VLMs) have achieved strong performance on general multimodal reasoning, yet remain challenged in integrating nonlocal visual information to support semantically underdetermined visual reasoning. We describe this challenge as Fragmented Visual Reasoning. To this end, we propose Credit Assignment for Visual Evidence (CAVE), a structured process-reward method based on GRPO for interleaved visual reasoning. Specifically, CAVE evaluates the contribution of intermediate steps at the action level via three complementary reasoning process signals: belief update, evidence acquisition, and adaptive focus control, thereby guiding the model to optimize each reasoning action and learn more reliable visual reasoning strategies. Meanwhile, we construct TRACER-Bench, which covers four nonlocal and semantically confusable reasoning dimensions and provides key intermediate evidence to supervise reasoning paths. Experiments demonstrate that CAVE substantially improves performance on tasks requiring fragmented visual evidence integration, covering both public benchmarks and our newly introduced TRACER-Bench, while retaining competitive performance on general multimodal evaluations. Further analyses reveal that CAVE effectively improves the visual reasoning capacity and exhibits stronger robustness under longer-range and deeper cross-region dependencies. Our code and the proposed benchmark are available at: https://anonymous.4open.science/r/Anonymous_code-2D8B/.