Attention
-
MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models
2026-10-16 05:20:45Khuyen Pham, Tran Nguyen Le, Edoardo Fazzari, Cesare StefaniniAbstract
Vision-language-action (VLA) models have shown strong promise for generalist robotic manipulation, but their inference cost remains a major bottleneck for real-time control. This limitation is especially pronounced in flow-matching VLAs such as $\pi_{0.5}$, where the action expert is evaluated repeatedly during iterative denoising. We introduce MambaAct, an efficient Mamba action expert for flow-matching VLA policies. MambaAct retains the pretrained visual-language prefix encoder of $\pi_{0.5}$, while replacing its transformer-based action expert with a compact Mamba-based action expert. Within this expert, Mamba serves as the action mixer that models dependencies across the noisy continuous-action sequence. To preserve visual-language conditioning without repeated token-level cross-attention, MambaAct uses a lightweight Context Compressor to map Vision-language model (VLM) prefix tokens into a compact conditioning signal that gates and modulates each Mamba action block. MambaAct uses a $10\times$ smaller action expert and reduces flow-matching denoising latency by $64.8\%$ on LIBERO and $58.8\%$ in real-world deployment, while improving mean LIBERO success by $1.08$ percentage points across three independent fine-tuning seeds and maintaining gains on LIBERO-Plus and real-world benchmarks. Our code and model weights are publicly available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22951
PDF
-
Many Biases, One Circuit: Post-Training Tunes a Pretrained Cue-Routing Mechanism
2026-10-16 05:00:12Prakhar Gupta, Terry Jingchen Zhang, Zhijing JinAbstract
Language models often abandon a correct answer when the prompt contains a cue, a hint toward a wrong option such as "I think the answer is B." Sycophancy and related biases show this failure. They are often studied separately and blamed on alignment. We test this view causally, tracing the attention heads behind these biases through the stages of two post-training pipelines. In multiple-choice tests on most models, four cue types rely largely on the same few heads, a shared circuit already present in pretrained models. Most base models follow cues in plain text. Fine-tuning a base model on about 200 generic assistant examples without cues activates these heads in chat format, so cue-following can be a side effect of assistant tuning. The final reinforcement learning (RL) stage strengthens the effect of cues in almost every comparison. In Tulu-3-8B and OLMo-2-7B, we swap the attention patterns of six middle layers between the RL checkpoint and its predecessor. For suggested-answer cues, this shifts each checkpoint's cue effect most of the way to the other's level. Post-training thus seems to change when and how strongly models use this circuit. Mitigation for these multiple-choice biases could target this circuit.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22937
PDF
-
Sep-Latent: Separator Tokens as Natural Compression Anchors for Implicit Reasoning
2026-10-16 04:39:25Yinlong Xu, Jianfeng Huang, Zhengdong Zhou, Mingjie Lin, Fuyu Lv, Dan Ou, Hongxia Xu, Jian WuAbstract
Large Language Models (LLMs) have demonstrated remarkable success in complex reasoning through explicit Chain-of-Thought (CoT), but the associated computational overhead and token verbosity remain significant barriers to efficiency. While implicit reasoning in continuous latent spaces offers a promising alternative, existing methods like Coconut and CODI often rely on fixed-length latent sequences and suffer from representational collapse. In this paper, we propose Sep-Latent, an adaptive implicit reasoning framework that utilizes separator tokens as structural semantic anchors. By introducing an Attention-Decay mechanism, we force the model to aggregate and compress reasoning logic into these anchors during training. We then align explicit and implicit pathways through multi-level distillation across all layers to ensure logical consistency. Crucially, Sep-Latent enables dynamic reasoning depth by adaptively terminating the latent thinking phase based on query complexity. Experimental results on the GSM8k-Aug benchmark show that Sep-Latent achieves an accuracy of 56.8\% with an average of only 5.4 tokens, significantly outperforming prior fixed-length implicit models in efficiency while narrowing the performance gap with explicit CoT. Further analysis via PCA and symbolic variant testing confirms that our method ensures semantic grounding and exhibits superior robustness to perturbations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22924
PDF
-
FederatedBytes: Tokenizer-Free, Byte-Level Small Language Models for Efficient Multimodal Federated Learning
2026-10-16 04:14:35Ali Sarabadani, Massimo Piccardi, Mehran Abolhasan, Farzad Tofigh, Ian Zhou, Justin LipmanAbstract
Federated Learning (FL) is a powerful machine learning framework that allows building a global model from the contributions of multiple, distributed clients, while ensuring the privacy of their local data. However, its heavy computational and communication overheads challenge its use for contemporary AI workloads such as natural language processing (NLP) and multimodal analysis. For this reason, this paper presents a novel small language model that leverages byte-level text and multimodal data processing, enhanced by a linear attention mechanism, to enable manageable Federated Learning of NLP and multimodal analysis tasks. The proposed model, nicknamed FederatedBytes, manages to process text and any other data types, including images, by treating its input as a mere sequence of bytes, while its lean linear attention mechanism allows abating the overall memory and computational footprint. To prove the effectiveness of the proposed model we have carried out an extensive performance evaluation on sentiment analysis tasks in multilingual and multimodal settings over the Multilingual Amazon Reviews and ScienceQA datasets. The experimental results show that FederatedBytes has been able to outperform comparable models such as ByT5 and LSTM, achieving higher accuracy while significantly reducing memory and communication overheads.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22906
PDF
-
ToPA: Block-wise Toeplitz Adaptation for Expressive and Efficient Fine-Tuning
2026-10-16 03:38:18Sicong Li, Qianqian Xu, Zhiyong Yang, Zitai Wang, Longtao Huang, Xiaochun Cao, Qingming HuangAbstract
Parameter-Efficient Fine-Tuning (PEFT) has emerged as a prevalent strategy for adapting large pre-trained models. Among PEFT techniques, reparameterization-based methods, particularly those that employ low-rank or sparsity structures, have attracted significant attention. However, these approaches often underperform full fine-tuning. Specifically, low-rank methods constrain the update rank, which is misaligned with the inherently high-rank nature of full fine-tuning. In contrast, sparsity-based techniques limit the flexibility of the parameter space and compromise connectivity across weights. To overcome these limitations, we explore the use of Toeplitz matrices, whose entries are constant along each diagonal, therefore providing a compact parameterization without imposing explicit rank or sparsity constraints. A straightforward approach is to update weights via a product of Toeplitz matrices. However, similar to RNNs, long multiplicative chains often lead to gradient instability. To address this issue, we propose \textbf{ToPA}, a more refined version of this idea that replaces standard Toeplitz matrices with block-wise Toeplitz, greatly increasing each factor's capacity and shortening the chain, thereby stabilizing training and improving expressiveness. We theoretically show that products of block-wise Toeplitz matrices can approximate arbitrary matrices under shorter chain length, justifying the structural design of ToPA. Furthermore, we demonstrate that ToPA offers greater expressivity than existing low-rank and sparse parameterizations. Empirical evaluations across 21 NLP and CV datasets, spanning 5 model architectures, consistently validate the effectiveness of ToPA.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22883
PDF
-
Measurement-Adaptive Light-Cone Transform for Robust Non-Line-of-Sight Imaging
2026-10-16 02:47:58Lianfang Wang, Xueying Liu, Junnan Dong, Qinghua Zhang, Jun Liu, Yong Wang, Yuping DuanAbstract
Practical non-line-of-sight (NLOS) measurements are affected by noise and deviations from idealized light transport, which can degrade reconstruction with conventional light-cone transform (LCT) methods. We propose the Measurement-Adaptive Light-Cone Transform (MALCT), a physics-guided framework that retains the efficient Fourier-domain inversion structure of LCT. {Within a unified variational formulation, MALCT combines the analytical response with a shared learned response through measurement-conditioned weighting and predicts a measurement-conditioned, frequency-dependent regularization map.} An attention-enhanced spatio-temporal feature extraction module captures long-range temporal dependencies and local spatial correlations to support reconstruction under low-signal conditions. Experiments on synthetic data and real measurements from different NLOS systems demonstrate improved reconstruction quality. Controlled evaluations further assess robustness to temporal-response broadening and reduced acquisition time. MALCT also preserves scene structures with spatial sampling grids down to $32\times32$, supporting its potential for NLOS imaging under limited acquisition budgets.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22847
PDF
-
Causal Edit Serialization
Abstract
Standard autoregressive language models generate text sequentially, but revision requires referring back to tokens in the input context or those already generated. Causal Edit Serialization (CES) is introduced as a framework for expressing such revisions within the same causal token stream. Special pointer tokens are paired with predicted source addresses, allowing insertion, deletion, replacement, copying and movement to be localized explicitly. A lightweight localization head and dual-index rotary position encoding support both address prediction and conditioning on previously emitted pointers, while retaining a standard causal attention mask. A pretraining procedure combines ordinary language modeling with serialized repairs derived from random corruptions, model-generated edits and repeated text. Across model scales and editing baselines, CES models demonstrate complete applied edits with modestly higher clean-text bits per byte than clean autoregressive models. Supervised and reinforcement-learning post-training provide a grammatical-error-correction proof of concept, producing more beneficial than harmful changes under reference-blind model judging. Correct repair programs also improve subsequent text prediction while the original context and its cached states are retained. CES provides a common interface for generation and localized revision, with potential extensions to iterative refinement and additional pointer-addressed transformations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22820
PDF
-
Causal Latent Workspaces for Looped Transformers
Abstract
Reusing a Transformer block across depth increases computation without replicating its parameters, but successive passes have no dedicated state through which positions can exchange compact summaries of their causal prefix. We introduce the causal latent workspace Transformer: an anchored, four-round shared-depth architecture in which each position writes gated vectors into prefix-specific latent slots and retrieves a content-addressed summary before the next shared block application. This path complements causal self-attention while the original input embedding remains accessible at every round. Across three byte-level corpora, three seeds, and two matched parameter budgets, the proposed model achieves the best mean held-out perplexity in five of six settings against five Transformer controls. At 101.71M parameters it leads on WikiText-103 raw, enwik8, and TinyStories, reaching perplexities 2.7820, 2.7843, and 1.6823; these improve upon the strongest non-workspace control on each corpus by 1.06%, 0.73%, and 1.01%. All nine seed-labelled comparisons at this budget favor the workspace. At 0.918M parameters it leads on WikiText-103 and TinyStories, whereas a conditioned loop is stronger on enwik8. Architectural ablations show that adding the workspace improves upon the anchored loop in all six settings and support refreshing its causal summaries at each round. Together, these results identify a causal, content-addressable prefix workspace as an effective structural augmentation of looped Transformers.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22796
PDF
-
Reading the News Without Hindsight: NoLBERT Topics and the Cross-Section of Stock Returns
2026-10-15 23:27:49Kevin Foley, Jonathan Hartadi, Shivesh Prakash, Charishma Takkallapalli, Swapnil VatsalAbstract
Language models that read financial news are usually trained on text from the same years used to test them, so a backtest can reward hindsight. We build news topics from NoLBERT, an encoder pretrained only on text published between 1976 and 1995, adapted into a sentence encoder by contrastive fine-tuning on passages of the same article, with no labels or returns. We compare it with LDA, MiniLM and FinBERT readers on 1.26 million articles from 2012 to 2024, convert each reader's topics into attention shocks and stock exposures, and forecast the returns of the 1,000 largest US stocks with eight models at three rebalancing frequencies. Models are fitted on 2013-2019, choices are made on 2020-2021, and 2022-2024 is held out. The experiment with the best validation score loses money at every frequency. Measuring attention shocks against the past 63 trading days instead of 5 raises the test rank IC of every reader. NoLBERT gives the most stable topics and forecasts about as well as LDA. An equal-weight combination of all readers and models earns a monthly net Sharpe ratio of 1.12, significant against zero before adjusting for the search but not against a price-only benchmark.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22745
PDF
-
LEMON: Equivariant Learning over Attention Maps for Zero-Shot Misbehavior Monitoring on New LLMs
2026-10-15 22:53:45Guy Bar-Shalom, Mor Geva, Fabrizio Frasca, Haggai MaronAbstract
Large language models exhibit misbehaviors that hinder deployment in high-stakes settings. One approach to monitor such failures is to use an external LLM as a monitor, but this is prohibitively expensive. A well-known and efficient class of monitors can detect such failures from internal representations (such as activation probes), but are tied to the model they were trained on, requiring costly re-annotation as new LLMs appear. A natural way to avoid this repeated annotation cost is to develop internal-representation based monitors that transfer zero-shot across LLMs. However, such transfer is challenging because internal representations are not naturally aligned across models; in particular, attention maps, known to contain signals of misbehavior, may differ in head function, ordering, and model depth. In this work, we characterize those cross-model attention mismatches as algebraic symmetries and use geometric deep learning principles to build LEMON (Language-model Equivariant MOnitoring Network) an online graph neural network-based monitor, whose predictions are invariant under these symmetries. LEMON processes attention maps token-by-token, as they are produced and outputs a score after every token (or every window of $W$ tokens) at negligible cost. Across 16 LLMs and six datasets, in the zero-shot cross-model setting, LEMON outperforms all baselines on both hallucination and jailbreak monitoring tasks, improving over the strongest by up to 14 ROC-AUC points.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22723
PDF
-
Risk and Refusal: Mechanisms of LLM Responses to Suicidal Ideation
2026-10-15 22:34:45Drishti Goel, Soorya Ram Shimgekar, Jeongah Lee, Violeta J. Rodriguez, Daniel S. Brown, Ravi Karkar, Koustuv SahaAbstract
Refusal is a common safety mechanism in large language models (LLMs), yet in contexts of suicidal ideation (SI), refusing to engage can itself constitute a safety failure. We evaluate how eight instruction-tuned LLMs respond to SI and self-harm disclosures across three datasets, varying whether models are informed that a disclosure expresses SI and whether they receive a clinician-developed response protocol, and complement this with mechanistic analyses in six open-weight models. Our results reveal a nuanced relationship between refusal and response-level risk. At baseline, elevated risk in model responses is driven primarily by inattention to the disclosure, alongside refusal to engage. Mechanistic analyses suggest that models may refuse SI disclosures through the same direction that mediates their refusals of harmful requests. However, informing models of the SI context, or suppressing refusal directly, increases engagement without reliably lowering response risk, and may raise it. In contrast, the clinician-guided protocol consistently reduces response-level risk while also reducing refusal. Together, these analyses and ablations motivate the need for models that refuse only when warranted and engage in interactionally safe ways when users disclose SI.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22715
PDF
-
First Passage Attention
Abstract
Augmenting softmax attention with forget gates typically relies on summing decay contributions along the key--query path. Because such accumulation is monotonic, these mechanisms cannot selectively erase intermediate context while preserving older, structurally relevant tokens. We propose First Passage Attention (FPA), which shifts from path summation to path extrema by lowering a key's logit proportionally to the deepest dip of a learned scalar field below that key's level over the intervening span. FPA encompasses the Forgetting Transformer (FoX) as a special case, and when the field and levels follow a shared learned walk, the alive keys with unpenalised logits naturally form a stack. We show that a constant-dimension FPA head realises any subset of keys left alive by subsequent dips, whereas softmax heads with positional biases or token-level forget gates require a head dimension linear in the number of keys. In synthetic experiments, one-layer FPA heads converge to these target sets while comparable softmax and FoX baselines fail on every seed. Scaling to a 110M-parameter language model, FPA matches FoX in perplexity and uniquely achieves perfect passkey retrieval at $8\times$ the training length, even with its key--value cache compressed to about $25\%$.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22714
PDF
-
Brain-Aligned Multimodal LLMs for Social and Emotional Understanding of Cartoon Movies
2026-10-15 22:22:03Yiran Huang, Zeyong Zhang, Bharat Biswal, Xin Di, Mengjia XuAbstract
Multimodal large language models (MLLMs) offer scalable annotations of social and affective content, but their temporal agreement with human judgments and neural predictive value require further evaluation. We compared diffusion-based and autoregressive MLLMs with continuous human Theory of Mind (ToM) and affective valence ratings across three short cartoon movies. Behavioral analyses assessed temporally aligned Pearson correlations, rating distributions, and the effects of preceding scene descriptions. Ratings from each source were separately used in subject-specific regression models to predict regional fMRI responses. Model ratings captured aspects of human temporal variation despite systematic distributional differences, with generally stronger alignment for valence than ToM. Preceding context improved alignment in some conditions, depending on the model and movie. In \emph{The Present}, more regions showed significant encoding for MLLM ratings than for human ratings, whereas effects were more limited in the other movies. Spatial overlap across rating sources occurred mainly among MLLMs; overlap between human and model ratings was limited and involved primarily visual, dorsal-attention, and posterior default-mode regions. These findings support MLLMs as scalable tools for generating brain-predictive annotations of social and emotional content, while highlighting systematic rating differences and model- and stimulus-dependent alignment with human judgments.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22706
PDF
-
LGGS: Longitudinal MRI Reconstruction via Guided Group Sparsity
2026-10-15 21:02:31Nikola Janju\v{s}evi\'{c}, Hersh Chandarana, Yao Wang, Li FengAbstract
Serial MRI is routinely performed in current clinical practice for disease surveillance and applications such as MR-guided radiotherapy. However, long MRI scan times increase costs and limit patient access, making accelerated image reconstruction of broad clinical importance. Previous scans of the same patient contain image information highly correlated with the current exam, yet existing reconstruction methods rarely exploit this information. Major challenges in the effective use of prior scans include inter-visit misregistration, pathological changes, and image phase variation. In this work, we propose Longitudinal Guided Group Sparsity (LGGS), an interpretable deep-learning framework for longitudinally informed MRI reconstruction derived from classical group sparsity. The prior scan is incorporated into the reconstruction regularizer through learned phase-invariant circulant attention that identifies corresponding features across exams. This formulation provides robustness to local in-plane and through-plane misregistration as well as inter-visit phase variation. Across acceleration rates, LGGS achieves state-of-the-art performance on longitudinal T1-weighted multi-coil brain MRI data, outperforming the evaluated unguided and longitudinally informed reconstruction methods, with substantially fewer learned parameters. We further evaluated the contributions of multiple guide slices and phase-invariant similarity and demonstrated that accelerated prior images can also serve as effective guides with minimal degradation in reconstruction quality.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22643
PDF
-
SAFETYEDIT: Sparse Internal Targets with Prompt-Wide Control for LLM Jailbreak Backdoors
2026-10-15 20:57:15Yutong Ke, Yansu Wang, Zeyu Zhang, Xihao Guo, Chongwen Zhao, Kaizhu HuangAbstract
Model-editing backdoors aim to induce harmful responses under a trigger while preserving ordinary behavior on untriggered input. Recent model-editing approaches target affirmative response prefixes through rank-one updates to a small subset of model weights. Yet when this prefix-level target is satisfied, the model may still revert to refusal or produce a coherent, apparently compliant response that does not address the request. To overcome this limitation, existing internal-representation approaches shift focus from output tokens to dense, high-dimensional compliance directions in latent space. Existing representation-level approaches rely on preservation constraints to maintain conditionality, motivating us to ask whether a smaller internal target can offer a different effectiveness-preservation trade-off. Motivated by the comparable control capability of single-coordinate interventions, we investigate a sparse refusal-related MLP coordinate as this target. We propose SAFETYEDIT, which optimizes a single refusal-related coordinate and uses KV-cache marking in early attention layers to propagate a prefix-trigger signal across subsequent instruction positions. On some checkpoints a downstream MLP reader converts this signal into control of the coordinate, while on others the propagated signal alone suffices; the method is defined by the sparse target rather than by a fixed writer-reader pair. Across four open-weight LLMs and three dataset groups, SAFETYEDIT achieves the highest triggered Guard ASR in 11 of 12 comparisons and the highest mean StrongREJECT score in all 12, relative to the baselines evaluated for each model. Crucially, comprehensive ablation and behavioral audits confirm generally minimal changes in untriggered behavior and general capabilities in the tested configurations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22640
PDF
-
TRACE: Trustworthy Reasoning from Acoustic Cues and Evidence
2026-10-15 19:59:23Giovana Morais, Yujin Kim, Atilay Kucukoglu, Magdalena FuentesAbstract
We introduce TRACE (Trustworthy Reasoning from Acoustic Cues and Evidence), a benchmark for diagnosing the reliability of large audio-language models (LALMs) by revealing when, and under what acoustic conditions these models behave untrustworthily. Existing benchmarks primarily measure whether a model produces the correct answer, but correctness alone cannot determine whether the answer is grounded in acoustic evidence or instead reflects language and contextual priors. TRACE evaluates model behavior along three complementary dimensions: (1) task performance, whether the answer is correct; (2) evidence alignment, whether the model is attending to the audio regions required to answer the question; and (3) robustness, whether answers and attributed evidence respond coherently to controlled changes such as event order, overlap, signal-to-noise ratio, or background content. TRACE targets temporal ordering, counting, and event existence, three tasks that remain challenging for current models. We synthesize contextually informed acoustic scenes that allow us to test whether models hallucinate events associated with the context but absent from the audio. This synthesis also allows us to obtain a scalable benchmark for which we know the answers by design. For each scene, we create controlled variants in a programmatic way to stress-test model robustness. We introduce attribution-based metrics that quantify model attention at scale on target events and non-target events. Evaluation across five state-of-the-art models shows that current LALMs often detect where relevant sound evidence occurs, but they do not reliably interpret it, integrate it into the answer, or respond to controlled changes in that evidence. Our benchmark also shows that models still struggle the most with ordering and counting questions.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26894
PDF
-
Gen4P: Video Generators are 4D Perceivers
2026-10-15 17:59:48Yitong Xia, Zeren Jiang, Chuanxia ZhengAbstract
We introduce Gen4P, a unified framework for recovering 4D geometry and motion from monocular video. We show that a pretrained video generator can be repurposed into a feed-forward 4D perceiver without iterative denoising. Building on on-demand querying, Gen4P processes clean video latents in a single encoder pass and predicts the requested 4D attributes through a lightweight cross-attention decoder. The on-demand query formulation enables flexible spatial sampling, and our proposed continuous-time queries further provide temporal flexibility. To strengthen geometric knowledge, we propose a straightforward plug-in solution by incorporating a pretrained video-depth LoRA as a frozen geometry prior, improving overall performance without adding extra trainable parameters. Gen4P achieves state-of-the-art or competitive performance on tasks of 3D point tracking, video depth and camera pose estimation. Qualitative results on in-the-wild videos illustrate its capability of tracking complex motion while maintaining the delicate structure. Together, these results demonstrate the potential of video generative pretraining for unified 4D perception. See this https URL for more visual results.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22371
PDF
-
Perceive, Predict, and Plan: a Unified World Action Model for Egocentric Dexterous Manipulation
2026-10-15 17:59:07Zirui Wu, Chen Guo, Tianjian Jiang, Zeren Jiang, Zeming Li, Lingchen Yang, Longteng Duan, Ying Xue, Handi Yin, Shenyuan Gao, Martin R. Oswald, Jun Zhang, Ping Tan, Jie SongAbstract
We introduce P$^{3}$-WAM, a unified world action model for egocentric dexterous manipulation that learns 3D perception, action-conditioned prediction, and goal-directed planning through three complementary modes: Perceive, which reconstructs articulated hand motion and camera trajectories from video; Predict, which generates visual futures under prescribed actions; and Plan, which jointly generates action trajectories and their anticipated visual consequences conditioned on an initial image, a language instruction, and an optional visual goal. Our model integrates a compact, structured 3D action representation into a pretrained video diffusion transformer through a mixture-of-transformers architecture with shared base weights, modality-specific adaptation, and bidirectional visual-action attention. Randomized modality visibility lets video and 3D actions serve as either conditioning context or prediction targets across the three modes. Experiments show that unified training improves all three capabilities over single-mode training, including gains in planning quality under a vision-language model evaluation. Through unified training, our model outperforms specialized baselines on 3D hand reconstruction and action-conditioned video prediction across four datasets, while producing goal-directed manipulation videos with stronger instruction adherence and action progression than video-generation baselines. Project Page: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2610.22363
PDF
-
DK-VLM: Geometry as Position for Spatial Reasoning in Vision-Language Models
2026-10-15 17:58:15Han Su, Meng Tian, Tianyu Huang, Mingyu Gao, Lennox Wei, Dongxu Li, Jianhua Han, Hang Xu, Xiaohe Wu, Wangmeng ZuoAbstract
Vision-language models still struggle to understand 3D spatial relations across views and update spatial judgments as the observer moves. A common approach fuses 3D reconstruction features into visual tokens, but aligning these features with pretrained semantic representations remains challenging under limited spatial supervision. We propose DK-VLM with Dual-Kernel Attention, which computes attention from both a visual token's input position and its geometric position in the scene. A semantic kernel retains the pretrained model's native attention and positional encoding, while a separate geometric kernel uses estimated 3D points and camera poses to directly modulate attention through point displacements and relative view transformations. Then, a learned gate fuses their outputs. Geometry thus enters attention as position rather than as fused reconstruction features. We further introduce Reference-Frame Composition (RC) supervision to train the model to update its reference frame across successive judgments. It provides stepwise turn and heading targets derived from verified single-step annotations. Among the compared methods, DK-VLM-8B achieves the highest average score across six spatial benchmarks and the highest overall scores on VSI-Bench (74.33) and ReVSI-32 (58.28). RC supervision further improves multi-step spatial reasoning, while the full model preserves general video understanding. Project page: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22356
PDF
-
Not Just a Subject: Capturing Adaptive Scene-Specific Variance for Visual Storytelling
2026-10-15 17:51:49SeungJu Cha, Ye-Chan Kim, Kwanyoung Lee, Dong-Jin KimAbstract
Recent advancements in diffusion models have significantly improved text-to-image generation, enabling their application to visual storytelling, where coherent image sequences are generated from narrative prompts. However, existing methods primarily focus on identity preservation and often fail to adapt to diverse scene variations, resulting in semantic misalignment and structural inflexibility. We propose EOS-Story, a training-free framework that harmonizes identity consistency with scene-specific adaptation across text and vision spaces. In the text space, we decouple the narrative into identity and scene sub-prompts to prevent information dilution, but this leaves the scene without access to appearance cues. To bridge this gap, we introduce Cross-Contextual Bridging, which softly injects a phrase-level appearance residual into the scene's <EOS> embedding. In the vision space, we find that scale mismatch between identity and scene subjects causes a few identity tokens to be repeatedly transferred, producing structural artifacts. We address this with Adaptive Identity Preservation, which anchors identity features at the scene-specific scale (Identity Anchoring) and integrates them via Identity-Aware Self-Attention. To evaluate performance under varied spatial compositions, we extend the Consistory+ benchmark with diverse framing scenarios (Consistory+_+). Extensive experiments show that EOS-Story achieves competitive or superior performance on both benchmarks.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22320
PDF
