GridProbe uses posterior probing on a KxK frame grid to adaptively select question-relevant frames, delivering up to 3.36x TFLOPs reduction with accuracy within 1.6 pp of the full-frame baseline on Video-MME-v2.
hub Mixed citations
Sigmoid loss for language image pre-training
Mixed citation behavior. Most common role is background (40%).
20
Pith papers citing it
Background
40% of classified citations
hub tools
citation-role summary
background 5
method 4
baseline 1
citation-polarity summary
representative citing papers
VEGA improves spatial reasoning in VLA models for robotics by aligning visual encoder features with 3D-supervised DINOv2 representations via a temporary projector and cosine similarity loss.
TrajGANR learns continuous neural representations of trajectories to enable fine-grained alignment with street-view images and locations in a joint multimodal self-supervised objective, outperforming prior geospatial MSSL methods on urban mobility and road tasks.
ST-BiBench reveals a coordination paradox in which MLLMs show strong high-level strategic reasoning yet fail at fine-grained 16-dimensional bimanual action synthesis and multi-stream fusion.
SkyNative introduces an encoder-free architecture using raw patch tokens and modality-specific parameters in a unified autoregressive model to improve image-grounded reasoning in remote sensing vision-language tasks.
Attention Hijacking is a new attack that improves cross-query transferability in VLMs by explicitly steering internal attention to a persistent image-dominant pattern.
LDDR proposes a linear DPP-based dynamic-resolution frame sampler that achieves 3x speedup and up to 2.5-point gains on video MLLM benchmarks by selecting non-redundant frames and allocating tokens accordingly.
HarmoWAM unifies predictive and reactive control in world action models via an adaptive gating mechanism to deliver improved zero-shot generalization and precision in robotic manipulation.
DeCIR improves projection-based zero-shot composed image retrieval by decoupling endpoint and semantic transition alignment with separate low-rank adapters merged by LRDM, showing gains on CIRR, CIRCO, FashionIQ, and GeneCIS.
Prior-Aligned AutoEncoders shape latent manifolds with spatial coherence, local continuity, and global semantics to improve latent diffusion, achieving SOTA gFID 1.03 on ImageNet 256x256 with up to 13x faster convergence.
Modality representations share dominant semantic geometry but have an anisotropic residual gap; AnisoAlign corrects source representations boundedly using target geometry for unpaired alignment.
TriRelVLA introduces triadic object-hand-task relational representations and a task-grounded graph transformer with a relational bottleneck to improve generalization in robotic manipulation across scenes, objects, and tasks.
Outlier tokens in DiTs are addressed with Dual-Stage Registers, which reduce artifacts and improve image generation on ImageNet and text-to-image tasks.
LaST-R1 introduces a RL post-training method called LAPO that optimizes latent Chain-of-Thought reasoning in vision-language-action models, yielding 99.9% success on LIBERO and up to 22.5% real-world gains.
LFRAG advances multimodal RAG to block-level retrieval with layout segmentation and cross-attention fusion, reporting SOTA retrieval, 7.20% higher answer accuracy, and 73.07% lower token consumption on the new LFDocQA benchmark.
SLQ adapts frozen MLLMs for multimodal retrieval by appending shared latent queries to text and image tokens and introduces KARR-Bench to test knowledge-aware reasoning retrieval.
SciTikZer-8B uses a new dataset, benchmark, and dual self-consistency RL to generate TikZ code for scientific graphics, outperforming much larger models like Gemini-2.5-Pro.
Emu3.5 is a native multimodal world model pre-trained on over 10 trillion vision-language tokens with next-token prediction, post-trained via reinforcement learning, and accelerated by Discrete Diffusion Adaptation for efficient interleaved generation and world exploration.
StableVLA adds an Information Bottleneck Adapter to VLA models that improves robustness to visual corruptions by 30% on average with under 10M extra parameters and no extra data, even when using a much smaller backbone.
ShellfishNet is a new benchmark of 8,691 images across 32 mollusc taxa for evaluating vision models on real-world underwater ecological monitoring tasks including robustness to degradation.
citing papers explorer
-
GridProbe: Posterior-Probing for Adaptive Test-Time Compute in Long-Video VLMs
GridProbe uses posterior probing on a KxK frame grid to adaptively select question-relevant frames, delivering up to 3.36x TFLOPs reduction with accuracy within 1.6 pp of the full-frame baseline on Video-MME-v2.
-
VEGA: Visual Encoder Grounding Alignment for Spatially-Aware Vision-Language-Action Models
VEGA improves spatial reasoning in VLA models for robotics by aligning visual encoder features with 3D-supervised DINOv2 representations via a temporary projector and cosine similarity loss.
-
TRAJGANR: Trajectory-Centric Urban Multimodal Learning via Geospatially Aligned Neural Representations
TrajGANR learns continuous neural representations of trajectories to enable fine-grained alignment with street-view images and locations in a joint multimodal self-supervised objective, outperforming prior geospatial MSSL methods on urban mobility and road tasks.
-
ST-BiBench: Benchmarking Multi-Stream Multimodal Coordination in Bimanual Embodied Tasks for MLLMs
ST-BiBench reveals a coordination paradox in which MLLMs show strong high-level strategic reasoning yet fail at fine-grained 16-dimensional bimanual action synthesis and multi-stream fusion.
-
SkyNative: A Native Multimodal Framework for Remote Sensing Visual Evidence Reasoning
SkyNative introduces an encoder-free architecture using raw patch tokens and modality-specific parameters in a unified autoregressive model to improve image-grounded reasoning in remote sensing vision-language tasks.
-
Attention Hijacking: Response Manipulation Across Queries in Vision-Language Models
Attention Hijacking is a new attack that improves cross-query transferability in VLMs by explicitly steering internal attention to a persistent image-dominant pattern.
-
LDDR: Linear-DPP-Based Dynamic-Resolution Frame Sampling for Video MLLMs
LDDR proposes a linear DPP-based dynamic-resolution frame sampler that achieves 3x speedup and up to 2.5-point gains on video MLLM benchmarks by selecting non-redundant frames and allocating tokens accordingly.
-
HarmoWAM: Harmonizing Generalizable and Precise Manipulation via Adaptive World Action Models
HarmoWAM unifies predictive and reactive control in world action models via an adaptive gating mechanism to deliver improved zero-shot generalization and precision in robotic manipulation.
-
Decoupling Endpoint and Semantic Transition Learning for Zero-Shot Composed Image Retrieval
DeCIR improves projection-based zero-shot composed image retrieval by decoupling endpoint and semantic transition alignment with separate low-rank adapters merged by LRDM, showing gains on CIRR, CIRCO, FashionIQ, and GeneCIS.
-
What Matters for Diffusion-Friendly Latent Manifold? Prior-Aligned Autoencoders for Latent Diffusion
Prior-Aligned AutoEncoders shape latent manifolds with spatial coherence, local continuity, and global semantics to improve latent diffusion, achieving SOTA gFID 1.03 on ImageNet 256x256 with up to 13x faster convergence.
-
Anisotropic Modality Align
Modality representations share dominant semantic geometry but have an anisotropic residual gap; AnisoAlign corrects source representations boundedly using target geometry for unpaired alignment.
-
TriRelVLA: Triadic Relational Structure for Generalizable Embodied Manipulation
TriRelVLA introduces triadic object-hand-task relational representations and a task-grounded graph transformer with a relational bottleneck to improve generalization in robotic manipulation across scenes, objects, and tasks.
-
Taming Outlier Tokens in Diffusion Transformers
Outlier tokens in DiTs are addressed with Dual-Stage Registers, which reduce artifacts and improve image generation on ImageNet and text-to-image tasks.
-
LaST-R1: Reinforcing Robotic Manipulation via Adaptive Physical Latent Reasoning
LaST-R1 introduces a RL post-training method called LAPO that optimizes latent Chain-of-Thought reasoning in vision-language-action models, yielding 99.9% success on LIBERO and up to 22.5% real-world gains.
-
LFRAG: Layout-oriented Fine-grained Retrieval-Augmented Generation on Multimodal Document Understanding
LFRAG advances multimodal RAG to block-level retrieval with layout segmentation and cross-attention fusion, reporting SOTA retrieval, 7.20% higher answer accuracy, and 73.07% lower token consumption on the new LFDocQA benchmark.
-
SLQ: Bridging Modalities via Shared Latent Queries for Retrieval with Frozen MLLMs
SLQ adapts frozen MLLMs for multimodal retrieval by appending shared latent queries to text and image tokens and introduces KARR-Bench to test knowledge-aware reasoning retrieval.
-
Scientific Graphics Program Synthesis via Dual Self-Consistency Reinforcement Learning
SciTikZer-8B uses a new dataset, benchmark, and dual self-consistency RL to generate TikZ code for scientific graphics, outperforming much larger models like Gemini-2.5-Pro.
-
Emu3.5: Native Multimodal Models are World Learners
Emu3.5 is a native multimodal world model pre-trained on over 10 trillion vision-language tokens with next-token prediction, post-trained via reinforcement learning, and accelerated by Discrete Diffusion Adaptation for efficient interleaved generation and world exploration.
-
StableVLA: Towards Robust Vision-Language-Action Models without Extra Data
StableVLA adds an Information Bottleneck Adapter to VLA models that improves robustness to visual corruptions by 30% on average with under 10M extra parameters and no extra data, even when using a much smaller backbone.
-
ShellfishNet: A Domain-Specific Benchmark for Visual Recognition of Marine Molluscs
ShellfishNet is a new benchmark of 8,691 images across 32 mollusc taxa for evaluating vision models on real-world underwater ecological monitoring tasks including robustness to degradation.