Every paper checked — nothing hidden
All papers
Every paper checked in the week of 12–18 Oct 2026 · Multimodal models — nothing is hidden. Brief papers first.
All1,924In brief3More (on-topic)553Not relevant1,333Not assessed19Older33Could not confirm19Blogs with no paper11All blogs13
| Paper | Topic | Relevance | Feedback |
|---|---|---|---|
| PlanPath: Exploration-Guided Orchestration for Patient-Level Pathology Reasoning The paper proposes an agent framework that plans multi-slide exploration and visual tool use for patient-level pathology visual question answering. Ranked #241New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Addresses multimodal visual question answering over complex medical pathology slides. | |
| Distilling Text LMs into Speech-to-Speech LMs by Matching Internal Text Predictions The paper proposes a knowledge distillation method that transfers text language model capabilities to speech-to-speech models using internal text interfaces. Ranked #242New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Builds unified speech-to-speech models transferring capabilities across text and audio modalities. | |
| Text-Only vs. Image-Aware VLM Judges for Manga Translation Evaluation The paper evaluates whether providing page images to vision-language models improves the detection of errors in Japanese-English manga translation. Ranked #243Surprising resultPeer-reviewed5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Analyzes vision-language models evaluating paired image-text contexts in manga translations. | |
| Lost in Contrast: Instruction-Preserving Decoding for Reliable Visual Question Answering This paper introduces an instruction-preserving decoding strategy to prevent visual contrastive decoding from discarding valid abstention responses. Ranked #244Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Improves visual question answering reliability in vision-language multimodal models via contrastive decoding. | |
| TAG: Calibrating SAM3 for Universal Image Segmentation without Dense Supervision Proposes a text anchor re-grounding method using low-rank adapters to improve concept recognition in Segment Anything Model 3. Ranked #245Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Aligns text anchors and visual features in SAM3 for open-vocabulary vision-language segmentation. | |
| MERIT: Multimodal Evidence Reasoning over Image Tables Proposes MERIT, a framework using visual table reasoning and adapted encoders to extract information directly from table images. Ranked #246New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Develops a multimodal framework for reasoning over table images and document structures. | |
| Multimodal Co-Generation With Continuous Bitstream Diffusion The paper applies continuous bitstream diffusion to joint and conditional generation across multimodal data without learned tokenizers. Ranked #247New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Proposes unified continuous bitstream diffusion for joint multimodal generation. | |
| A Persian Vision-Language-Action Dataset and Benchmark for Ambiguity- and Infeasibility-Aware Instruction-to-Plan Grounding The paper introduces a Persian vision-language-action dataset and benchmark to evaluate robotic instruction grounding under deliberate ambiguity and infeasibility. Ranked #248New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a vision-language dataset linking images and instructions to plans. | |
| LAPT-Scale: Scaling Embodied Policies with Latent Action Pre-Training on Human Videos Pre-trains embodied robot policies on human egocentric videos using learned continuous latent action representations. Ranked #249New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Develops a vision-language-action (VLA) policy combining video, semantics, and physical robot action prediction. | |
| Temporal Specification Guidance for Text-to-Video Generation The paper introduces an inference-time guidance method that enforces temporal logic constraints during text-to-video generation. Ranked #250New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Guides text-to-video diffusion generation to satisfy temporal specifications. | |
| VINCIE-NExT: Unlocking Video Editing from Images via In-Context Modeling Transfers editing capabilities from image pairs to video diffusion models using in-context visual demonstrations across decomposed sub-tasks. Ranked #251New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Presents a multimodal image-and-video diffusion architecture linking text instructions with visual frames. | |
| Beyond Visual Enhancement: Adaptive Multi-Context Steering to Mitigate LVLM Hallucinations Proposes an adaptive multi-source steering decoding method to mitigate hallucinations in vision-language models by balancing visual and textual context contributions. Ranked #252Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Directly investigates and improves vision-language model decoding across multimodal contexts. | |
| From Suppression to Repair: Mitigating Object Hallucination in Large Vision-Language Models via Localized Distribution Alignment Introduces a training-free inference method using Gaussian optimal transport to reduce object hallucination in vision-language models without suppressing valid representations. Ranked #253Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Improves representation faithfulness and generation quality in large vision-language models. | |
| What Do Audio-Language Judges of Generated Music Listen For? Four Probes with Controls from Expert Ratings Designs diagnostic probes to determine whether audio-language evaluation models judge music based on audio or prompt content. Ranked #254New testPreprint5/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Examines audio-language multimodal models and their ability to interpret paired audio and text. | |
| FastBench: Can Streaming VLMs Perceive High-Dynamic Real-World Streams? Introduces a benchmark evaluating streaming vision-language models on fast-moving, high-dynamic events in real-world video streams. Ranked #255New testPreprint5/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Focuses specifically on streaming vision-language models for continuous video understanding. | |
| HiPAD: A Unified Gait, Audio-Visual, and Personality Fusion Network for Depression Detection, with a Hierarchical Personality Ablation and the Calibration Limits of Domain-Adapted LLMs This paper presents a multimodal fusion network for depression detection combining gait, audio-visual data, and textual personality descriptions. Ranked #256Surprising resultPeer-reviewed5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Fuses text personality descriptions with audio, visual, and gait modalities. | |
| ChemExplain: A Multimodal Benchmark for Explanatory Chemical Reasoning Presents a multimodal benchmark testing whether models can explain observed chemical effects by linking structural features to reaction mechanisms. Ranked #257New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Evaluates multimodal models on understanding molecular and reaction images. | |
| Segment Where You Look: Attention as the Interface Between Language Models and Segmenters The paper uses attention maps from vision-language models to directly guide pixel decoders for referring image segmentation. Ranked #258Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Investigates attention mechanisms connecting vision-language models with segmentation decoders for text-guided segmentation. | |
| EmoGenie: Benchmarking and Evaluating Expressed and Evoked Emotion in T2I Generation with MLLMs Introduces EmoGenie, a benchmark evaluating expressed and evoked emotions in text-to-image generation and multimodal language model understanding. Ranked #259New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Introduces a benchmark evaluating multimodal LLMs on understanding and generating emotional visual content. | |
| LongForcing: Scaling Temporal Manifold Matching for Long-Horizon World Rollout Proposes LongForcing, a temporal manifold matching distillation approach that scales long-horizon video rollout stability. Ranked #260New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Focuses on scaling and improving temporal consistency in video diffusion foundation models. | |
| Benchmarking Multimodal Large Language Models for Zero-Shot Human-Object Interaction Benchmarks multimodal large language models paired with object detectors for zero-shot human-object interaction recognition. Ranked #261New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Benchmarks multimodal LLMs for zero-shot human-object interaction reasoning. | |
| FlowCred: Hierarchical Credit Assignment and Functional Coordination for Joint Text-to-Video Post-Training Proposes FlowCred, a hierarchical credit assignment framework to coordinate prompt optimization and video generator post-training. Ranked #262New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a post-training framework combining text prompts and video generation. | |
| LoopWAM: Closed-Loop Generalist Policy Improvement from Deployment Rollouts This paper introduces a closed-loop world-action model that improves generalist robot policies using deployment rollouts of mixed quality. Ranked #263New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Applies multimodal world-action models integrating video and action representations for robotics. | |
| Plan-Bench: Can Existing VLMs Go Well Beyond Perception to Reason and Plan in Embodied Worlds? This paper introduces a benchmark evaluating high-level reasoning and multi-step planning in vision-language models for embodied tasks. Ranked #264New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Evaluates high-level reasoning and multi-step planning capabilities across 28 vision-language models. | |
| Game2Real-Bench: Do Toy Games Reveal and Transfer to Real-World Visual Reasoning? This paper introduces a benchmark to analyze whether visual reasoning skills evaluated in toy games transfer to real-world tasks. Ranked #265Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Benchmarks vision-language models on spatial and constraint reasoning using game and real-world tasks. | |
| When, Where, and Why: Hierarchical Causal Explainability for Video Face Understanding with MLLMs Introduces an explainability framework and weakly supervised video-text corpus to evaluate causal and temporal facial expression understanding in multimodal models. Ranked #266New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Introduces a video-text corpus and causal concept framework for explainable facial video understanding in multimodal LLMs. | |
| MemAct: Streaming Spatial Understanding with Memory and Active Exploration Presents an embodied agent framework combining streaming spatial memory with active exploration for visual reasoning in physical environments. Ranked #267New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a multimodal streaming spatial understanding framework integrating vision-language reasoning with memory and active exploration. | |
| MicroCLIP: Connecting Text and Microscopy Images via Contrastive Language--Image Pretraining This paper introduces a vision-language model trained on paired microscopy image-text data for open-set biological image classification and retrieval. Ranked #268New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Connects biological microscopy images with natural language via contrastive pretraining. | |
| DVTR-KD: Distilling Vision-Language Models via Intermediate Visual Token Reconstruction This paper presents a knowledge distillation framework that reconstructs dense teacher visual tokens to train efficient compressed student vision-language models. Ranked #269New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Proposes a visual token reconstruction distillation method to make multimodal large language models more efficient. | |
| EgoProbe: Benchmarking Long-Range and Exact Memory in Streaming Egocentric Videos EgoProbe benchmarks long-range retention and event-specific recall under interference for streaming egocentric video models. Ranked #270New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Introduces a benchmark evaluating long-range multimodal memory in streaming egocentric videos. | |
| DIGS: Distribution-Informed Gaussian Splatting for Training-Free Open-Vocabulary 3D Segmentation Introduces a training-free open-vocabulary 3D segmentation approach that maintains per-primitive label distributions for clean object boundaries. Ranked #271New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Connects 3D Gaussian representations with multimodal vision-language embeddings. | |
| Meaningful Moments: A Corpus of VLM-Derived Temporal Importance Labels for Video Action Recognition Introduces a corpus of vision-language model generated importance scores to improve temporal sampling in video action recognition. Ranked #272New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Leverages vision-language models to score temporal importance across large video datasets. | |
| FoveaTree: Information-Density Trees for Query-Guided Visual Token Pruning in Video Large Language Models The paper introduces a training-free visual token pruning framework using information-density trees to speed up video large language model inference. Ranked #273New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Focuses on efficient token pruning for multimodal video large language models. | |
| VideoSlate: Stateful Commit Control for Temporally Grounded Agentic Video Reasoning The paper introduces a stateful commit-control framework with an evidence blackboard to manage revisions during agentic video reasoning. Ranked #274New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a stateful commit control framework for video grounding and question answering. | |
| Anomaly-Guided Observation and Two-Level Semantic Feedback for Training-Free Video Anomaly Analysis The paper presents a training-free semantic feedback framework using vision-language models to iteratively detect and explain video anomalies. Ranked #275New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Uses vision-language models for video anomaly understanding and semantic feedback. | |
| Semantically Steerable Image Tokenization The paper proposes an adaptive image tokenizer that balances pixel reconstruction and semantic features to prevent discarding critical semantic content during compression. Ranked #276New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Develops semantically steerable vision tokenization relevant to vision-language and multimodal model architectures. | |
| Semantic Saliency from Multi-Modal Large Language Model Scene Understanding Maps The paper develops an automated method that measures semantic saliency by evaluating how removing individual objects alters multimodal model scene descriptions. Ranked #277New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Uses multimodal large language models to compute semantic saliency and scene comprehension. | |
| EmoTrace-Bench: Diagnosing Emotional Expression in Text-to-Video Generation through Controlled Cues and Intensity The paper introduces a benchmark to evaluate text-to-video models across controlled emotional expression levels and specific audiovisual affective cues. Ranked #278New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Evaluates emotional expression and audiovisual cues in text-to-video multimodal models. | |
| SpikeOcc: Towards Resilient Multimodal 3D Occupancy Perception via Spiking Experts This paper presents a spiking mixture-of-experts architecture to improve the resilience of multimodal 3D occupancy perception against missing sensor inputs. Ranked #279New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a multimodal perception framework fusing RGB, LiDAR, and spiking experts. | |
| TartanMatch: Towards Universal Dense Correspondence Across Modalities This paper presents a unified model for dense visual correspondence across 25 pairs of multi-modal sensor inputs. Ranked #280Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Learns universal dense correspondence across diverse sensor modalities including RGB, thermal, and depth. | |
| How to Build a Benchmark That Actually Measures Taste This paper introduces a pairwise evaluation benchmark and protocol designed to measure subjective aesthetic taste while controlling for shortcut artifacts. Ranked #281New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Evaluates vision-language models on multimodal taste and craft judgment across visual items. | |
| Thinking with Images without Coordinates This paper introduces a method for multimodal models to reason over images using token-level latent regions instead of bounding-box coordinates. Ranked #282New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Introduces a coordinate-free visual reasoning method for multimodal large language models. | |
| Token-Budgeted Candidate Retrieval for 3D Scene-Grounded VLMs: Category Names vs. Pairwise Geometry This paper evaluates category name matching versus pairwise geometry metrics for candidate retrieval in context-constrained 3D vision-language models. Ranked #283Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Investigates token-budgeted candidate retrieval for 3D scene-grounded vision-language models. | |
| TIDE: Trajectory-Informed Skill Distillation for Efficient Text-to-Video Prompt Optimization The paper introduces a framework that distills iterative prompt search into rule modules for efficient text-to-video generation. Ranked #284New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Optimizes text-to-video prompts across multimodal generative models. | |
| BridgeWAM: Bridging Video and Action Experts via Latent Bridge Queries The paper presents an architecture that decouples video and action prediction models using latent bridge queries. Ranked #285New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Integrates visual, language, and proprioceptive modalities in a unified world-action model. | |
| DK-VLM: Geometry as Position for Spatial Reasoning in Vision-Language Models Presents a dual-kernel attention mechanism that uses 3D geometric positions to improve spatial reasoning in vision-language models. Ranked #286New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 17 Oct, 06:18 | On-topic Incorporates geometric positional attention into vision-language models for cross-view spatial reasoning. | |
| AffectWorld: Benchmarking Emotion-Driven Future World Prediction AffectWorld benchmarks world models on predicting emotion-driven human actions and subsequent environment changes across annotated video samples. Ranked #287New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Introduces a multimodal video benchmark evaluating emotion-action-state physical prediction. | |
| Heading Hierarchy Reconstruction for Complex and Long Documents with Level-Wise Multimodal Memory The authors present a framework using level-wise multimodal memory to reconstruct coherent heading hierarchies across long, multi-page documents. Ranked #288New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Uses vision-language models and multimodal memory to reconstruct document heading hierarchies. | |
| Know-Thy-Not: Training-Free Targeted Negation in Dual-Encoder Vision-Language Models The paper introduces a training-free technique and benchmark to help vision-language models correctly process and reject negated concepts in text prompts. Ranked #289New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Focuses on improving negation capabilities in dual-encoder vision-language models. | |
| D2: High-Capability Full-Duplex Spoken Language Models from Half-Duplex Backbones Converts half-duplex speech models into streaming full-duplex conversational models capable of simultaneous listening and speaking. Ranked #290New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Develops full-duplex spoken language models processing speech and text. | |
| REEL: Learning to Retrieve, Evaluate, and Evolve Visual Memory for Any Long-Horizon Multi-Shot Video Generation Introduces a general visual memory planner that retrieves, creates, and updates visual evidence across shots for long-horizon video generation. Ranked #291New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Focuses on visual memory planning and multi-shot video generation across multimodal inputs. | |
| Across Viewpoints and Time: Trajectory-Generalizable Physical Attacks on World Models and Video Understanding Models The paper develops an attack framework to generate physical adversarial textures that deceive video models across varying trajectories and viewpoints. Ranked #292New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Studies cross-viewpoint adversarial robustness for dynamic video understanding and world models. | |
| Flow-Support-Weighted Proposals for Test-Time Verified VLA Policies Introduces a training-time sampling method to prevent action proposal collapse in vision-language-action models tested with verifiers. Ranked #293New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Investigates action proposal scaling and flow decoders in vision-language-action models. | |
| Speculative Sampling in Interactive Video Generation This paper proposes Speculative Sampling to combine large bidirectional sampling windows with low-latency interaction in interactive video generation. Ranked #294Surprising resultPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Addresses sampling and latency for interactive text/conditioned video generation. | |
| What Matters for World Modeling in Vision-Language-Action Models? Evaluates four world-model integration paradigms to identify what improves vision-language-action policies in robotic tasks. Ranked #295New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Systematically analyzes world model integrations within Vision-Language-Action multimodal architectures. | |
| SaTPO: Text-to-Image Safety Alignment via Transition Preference Optimization The paper proposes a step-wise alignment method to suppress harmful content in text-to-image models while maintaining generation semantics. Ranked #296New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Studies safety alignment and preference optimization in multimodal text-to-image diffusion models. | |
| StateCraft: Augmenting Video World Models with Abstract-State Reasoning This paper pairs language models with video diffusion to maintain persistent abstract states across dynamic video world interactions. Ranked #297New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Combines language models with video diffusion to model dynamic visual environments. | |
| VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding Harness The paper examines failure modes in recursively self-improving execution harnesses for agentic video understanding models. Ranked #298New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Improves vision-language model harnesses for multimodal video understanding tasks. | |
| Frame Evidence:A Benchmark for Reference Frames and Visual Evidence in Multi-Image Spatial Reasoning The paper introduces a spatial reasoning benchmark to evaluate how models handle varying reference frames and altered visual evidence. Ranked #299New testPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Evaluates multi-image visual-language spatial reasoning and grounding. | |
| AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents This paper introduces AdaRSI, an on-demand recursive self-improvement framework for embodied agents that isolates and fine-tunes only persistent sub-step failures. Ranked #300New way to see a problemPreprint5/10 suggested | Multimodal models via arxiv · 15 Oct, 15:06 | On-topic Improves vision-language-action models for multimodal perception and robotic action execution. |
241–300 of 1,924← PreviousNext →