FeynmanBench is the first benchmark for evaluating multimodal LLMs on diagrammatic reasoning with Feynman diagrams, revealing systematic failures in enforcing physical constraints and global topology.
Canonical reference
Retrieval-Augmented Embodied Agents
Canonical reference. 100% of citing Pith papers cite this work as background.
90
Pith papers citing it
Background
100% of classified citations
citation-role summary
background 6
citation-polarity summary
co-cited works
years
2026 90roles
background 3polarities
background 3representative citing papers
ASH reaches 11.2/12 milestones in Pokemon Emerald and 9.9/12 in Zelda by self-improving via an IDM trained on its own trajectories to label internet video, while baselines plateau at roughly 6/12.
PG-OT builds prompt-specific Pareto frontiers and applies distribution-aware optimal transport to improve multi-reward alignment while introducing JDR and JCR metrics to measure synergy and hacking.
Vector Scaffolding uses Interior Gradient Aggregation, Progressive Stratification, and Rapid Inflation Scheduling to achieve 2.5x faster optimization and up to 1.4 dB higher PSNR in differentiable vectorization.
BICR uses blind-image contrastive ranking on frozen LVLM hidden states to train a lightweight probe that penalizes confidence on blacked-out inputs, yielding top calibration and discrimination across five models and multiple tasks at low parameter cost.
MulTaBench is a new collection of 40 image-tabular and text-tabular datasets designed to test target-aware representation tuning in multimodal tabular models.
GRCA uses emitter-centric geometric culling of rays per triangle to accelerate LiDAR simulation in arbitrarily dynamic scenes, reporting up to 14.55x speedup over Embree and 7.97x over OptiX.
LMMs perceive videos but underexploit visual content for causal reasoning due to textual shortcuts; ProCauEval diagnoses this and ADPO training reduces reliance on priors.
FLiD is a field-localized forgery detection method for identity documents that outperforms full-document baselines and general detectors with significantly fewer parameters.
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
ChartREG++ creates a new multi-target chart grounding benchmark with diverse cues and a code-driven synthesis pipeline for accurate masks, yielding a model that outperforms baselines and generalizes to real ChartQA charts.
AniMatrix generates anime videos by structuring artistic production rules into a controllable taxonomy and training the model to prioritize those rules over physical realism, achieving top scores from professional animators on prompt understanding and artistic motion.
Text-guided class-agnostic counting models exhibit significant weaknesses in grounding textual prompts to visual objects, as demonstrated by new negative-label and distractor tests on a multi-category dataset.
MIRL uses mutual information to guide trajectory selection and provide separate rewards for visual perception in RLVR for VLMs, achieving 70.22% average accuracy with 25% fewer full trajectories.
CSGuard binds diffusion-model watermarks to a secret matrix via compressed sensing, cutting forgery attack success from 100% to 28.12% while preserving 100% detection on legitimate images.
ESARBench is the first unified benchmark for MLLM-driven UAV agents that must explore, locate clues, and decide on victim positions in photorealistic simulated SAR environments.
FieryGS integrates LLM-based material reasoning, volumetric combustion simulation, and a unified renderer with 3D Gaussian Splatting to generate physically plausible and user-controllable fire in in-the-wild scenes.
SportsTime benchmark and CoTR method improve multimodal AI's temporal compositional reasoning and evidence grounding in long-form sports videos.
PolyChartQA is a new mid-scale dataset for multi-chart question answering that reveals a 27.4% accuracy drop for multimodal models on human-authored questions compared to AI-generated ones, plus a modest gain from a proposed prompting method.
HumanScore defines six metrics for kinematic plausibility, temporal stability, and biomechanical consistency to benchmark human motions in videos from thirteen state-of-the-art generation models, revealing gaps between visual appeal and physical fidelity.
DHCNet improves ultra-fine-grained visual categorization by progressively building holistic cognition from local discrepancies using self-shuffling and refinement on limited data.
SARR modifies trigonometric rotation encodings with object symmetry orders to produce unique continuous poses, enabling standard CNNs to outperform existing methods on symmetry-aware 6D pose estimation without custom losses or 3D models.
A survey that groups efficient video diffusion methods into four paradigms—step distillation, efficient attention, model compression, and cache/trajectory optimization—and outlines open challenges for practical use.
STOMP extends direct preference optimization to the multi-objective setting via smooth Tchebysheff scalarization and standardization of observed rewards, achieving highest hypervolume in eight of nine protein engineering evaluations.
citing papers explorer
-
FeynmanBench: Benchmarking Multimodal LLMs on Diagrammatic Physics Reasoning
FeynmanBench is the first benchmark for evaluating multimodal LLMs on diagrammatic reasoning with Feynman diagrams, revealing systematic failures in enforcing physical constraints and global topology.
-
ASH: Agents that Self-Hone via Embodied Learning
ASH reaches 11.2/12 milestones in Pokemon Emerald and 9.9/12 in Zelda by self-improving via an IDM trained on its own trajectories to label internet video, while baselines plateau at roughly 6/12.
-
Pareto-Guided Optimal Transport for Multi-Reward Alignment
PG-OT builds prompt-specific Pareto frontiers and applies distribution-aware optimal transport to improve multi-reward alignment while introducing JDR and JCR metrics to measure synergy and hacking.
-
Vector Scaffolding: Inter-Scale Orchestration for Differentiable Image Vectorization
Vector Scaffolding uses Interior Gradient Aggregation, Progressive Stratification, and Rapid Inflation Scheduling to achieve 2.5x faster optimization and up to 1.4 dB higher PSNR in differentiable vectorization.
-
Grounded or Guessing? LVLM Confidence Estimation via Blind-Image Contrastive Ranking
BICR uses blind-image contrastive ranking on frozen LVLM hidden states to train a lightweight probe that penalizes confidence on blacked-out inputs, yielding top calibration and discrimination across five models and multiple tasks at low parameter cost.
-
MulTaBench: Benchmarking Multimodal Tabular Learning with Text and Image
MulTaBench is a new collection of 40 image-tabular and text-tabular datasets designed to test target-aware representation tuning in multimodal tabular models.
-
Geometrically Approximated Modeling for Emitter-Centric Ray-Triangle Filtering in Arbitrarily Dynamic LiDAR Simulation
GRCA uses emitter-centric geometric culling of rays per triangle to accelerate LiDAR simulation in arbitrarily dynamic scenes, reporting up to 14.55x speedup over Embree and 7.97x over OptiX.
-
Perception Without Engagement: Dissecting the Causal Discovery Deficit in LMMs
LMMs perceive videos but underexploit visual content for causal reasoning due to textual shortcuts; ProCauEval diagnoses this and ADPO training reduces reliance on priors.
-
Field-Localized Forgery Detection for Digital Identity Documents
FLiD is a field-localized forgery detection method for identity documents that outperforms full-document baselines and general detectors with significantly fewer parameters.
-
The Moltbook Files: A Harmless Slopocalypse or Humanity's Last Experiment
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
-
ChartREG++: Towards Benchmarking and Improving Chart Referring Expression Grounding under Diverse referring clues and Multi-Target Referring
ChartREG++ creates a new multi-target chart grounding benchmark with diverse cues and a code-driven synthesis pipeline for accurate masks, yielding a model that outperforms baselines and generalizes to real ChartQA charts.
-
AniMatrix: An Anime Video Generation Model that Thinks in Art, Not Physics
AniMatrix generates anime videos by structuring artistic production rules into a controllable taxonomy and training the model to prioritize those rules over physical realism, achieving top scores from professional animators on prompt understanding and artistic motion.
-
Does it Really Count? Assessing Semantic Grounding in Text-Guided Class-Agnostic Counting
Text-guided class-agnostic counting models exhibit significant weaknesses in grounding textual prompts to visual objects, as demonstrated by new negative-label and distractor tests on a multi-category dataset.
-
MIRL: Mutual Information-Guided Reinforcement Learning for Vision-Language Models
MIRL uses mutual information to guide trajectory selection and provide separate rewards for visual perception in RLVR for VLMs, achieving 70.22% average accuracy with 25% fewer full trajectories.
-
CSGuard: Toward Forgery-Resistant Watermarking in Diffusion Models via Compressed Sensing Constraint
CSGuard binds diffusion-model watermarks to a secret matrix via compressed sensing, cutting forgery attack success from 100% to 28.12% while preserving 100% detection on legitimate images.
-
ESARBench: A Benchmark for Agentic UAV Embodied Search and Rescue
ESARBench is the first unified benchmark for MLLM-driven UAV agents that must explore, locate clues, and decide on victim positions in photorealistic simulated SAR environments.
-
FieryGS: In-the-Wild Fire Synthesis with Physics-Integrated Gaussian Splatting
FieryGS integrates LLM-based material reasoning, volumetric combustion simulation, and a unified renderer with 3D Gaussian Splatting to generate physically plausible and user-controllable fire in in-the-wild scenes.
-
Towards Temporal Compositional Reasoning in Long-Form Sports Videos
SportsTime benchmark and CoTR method improve multimodal AI's temporal compositional reasoning and evidence grounding in long-form sports videos.
-
Beyond Single Plots: A Benchmark for Question Answering on Multi-Charts
PolyChartQA is a new mid-scale dataset for multi-chart question answering that reveals a 27.4% accuracy drop for multimodal models on human-authored questions compared to AI-generated ones, plus a modest gain from a proposed prompting method.
-
HumanScore: Benchmarking Human Motions in Generated Videos
HumanScore defines six metrics for kinematic plausibility, temporal stability, and biomechanical consistency to benchmark human motions in videos from thirteen state-of-the-art generation models, revealing gaps between visual appeal and physical fidelity.
-
Divide-and-Conquer Approach to Holistic Cognition in High-Similarity Contexts with Limited Data
DHCNet improves ultra-fine-grained visual categorization by progressively building holistic cognition from local discrepancies using self-shuffling and refinement on limited data.
-
Towards Symmetry-sensitive Pose Estimation: A Rotation Representation for Symmetric Object Classes
SARR modifies trigonometric rotation encodings with object symmetry orders to produce unique continuous poses, enabling standard CNNs to outperform existing methods on symmetry-aware 6D pose estimation without custom losses or 3D models.
-
Efficient Video Diffusion Models: Advancements and Challenges
A survey that groups efficient video diffusion methods into four paradigms—step distillation, efficient attention, model compression, and cache/trajectory optimization—and outlines open challenges for practical use.
-
Pareto-Optimal Offline Reinforcement Learning via Smooth Tchebysheff Scalarization
STOMP extends direct preference optimization to the multi-objective setting via smooth Tchebysheff scalarization and standardization of observed rewards, achieving highest hypervolume in eight of nine protein engineering evaluations.
-
DPC-VQA: Decoupling Quality Perception and Residual Calibration for Video Quality Assessment
DPC-VQA decouples a frozen MLLM perceptual prior from a lightweight residual calibration branch to adapt video quality assessment to new scenarios with under 2% trainable parameters and 20% of typical MOS labels.
-
Beyond Perception Errors: Semantic Fixation in Large Vision-Language Models
VLMs display semantic fixation, with higher accuracy on standard rule mappings than inverse ones across 14 models, narrowed by neutral prompts but widened by loaded ones and affected by post-training alignment.
-
Mosaic: Cross-Modal Clustering for Efficient Video Understanding
Mosaic uses cross-modal clusters as the unit for KVCache organization in VLMs to achieve up to 1.38x speedup in streaming long-video understanding.
-
Choose, Don't Label: Multiple-Choice Query Synthesis for Program Disambiguation
Multiple-choice queries synthesized from Hoare triples enable more reliable identification of intended programs than labeled-example supervision in active learning for program disambiguation.
-
DiV-INR: Extreme Low-Bitrate Diffusion Video Compression with INR Conditioning
DiV-INR integrates implicit neural representations as conditioning signals for diffusion models to achieve better perceptual quality than HEVC, VVC, and prior neural codecs at extremely low bitrates under 0.05 bpp.
-
DetailVerifyBench: A Benchmark for Dense Hallucination Localization in Long Image Captions
DetailVerifyBench supplies 1,000 images and densely annotated long captions to evaluate precise hallucination localization in multimodal large language models.
-
A global dataset of continuous urban dashcam driving
CROWD is a new global dataset of 51,753 continuous urban dashcam segments spanning over 20,000 hours from 238 countries, with manual labels and automated object detections for routine driving analysis.
-
Routine Computing: A Systematic Review of Sensing Daily Life Dimensions Towards Human-Centered Goals
The first systematic review of routine computing synthesizes literature into a taxonomy of temporal, behavioral, cognitive, and variability aspects, outlining applications in health, accessibility, and adaptive support along with persistent challenges.
-
Sparse Code Uplifting for Efficient 3D Language Gaussian Splatting
SCOUP decouples 2D sparse code learning from 3D Gaussian optimization to deliver up to 400x training speedup and 3x better memory efficiency while matching accuracy on open-vocabulary 3D queries.
-
Rescaled Asynchronous SGD: Optimal Distributed Optimization under Data and System Heterogeneity
Rescaled ASGD recovers convergence to the true global objective by rescaling worker stepsizes proportional to computation times, matching the known time lower bound in the leading term under non-convex smoothness and bounded heterogeneity.
-
GRACE: Gradient-aligned Reasoning Data Curation for Efficient Post-training
GRACE scores reasoning steps via gradient alignment and trajectory consistency to select data subsets that match full performance with 5% of the data on Qwen3-VL-2B-Instruct.
-
EDGER: EDge-Guided with HEatmap Refinement for Generalizable Image Forgery Localization
A dual-branch system using frequency edge cues and CLIP-based synthetic patch detection for accurate, resolution-independent image forgery localization.
-
TB-AVA: Text as a Semantic Bridge for Audio-Visual Parameter Efficient Finetuning
TB-AVA uses text-mediated gated semantic modulation to enable efficient audio-visual alignment, achieving state-of-the-art results on AVE, AVS, and AVVP benchmarks.
-
PG-3DGS: Optimizing 3D Gaussian Splatting to Satisfy Physics Objectives
PG-3DGS couples 3D Gaussian Splatting with differentiable physics so that optimized shapes satisfy both visual fidelity and physical objectives such as pouring and aerodynamic lift, with real-world 3D-printed validation.
-
Pixal3D: Pixel-Aligned 3D Generation from Images
Pixal3D performs pixel-aligned 3D generation from images via back-projected multi-scale feature volumes, achieving fidelity close to reconstruction while supporting multi-view and scene synthesis.
-
Geospatial-Temporal Sensemaking of Remote Sensing Activity Detections with Multimodal Large Language Model
Introduces the SMART-HC-VQA dataset with 65k single-image and 2.3M temporal VQA examples plus an adapted LLaVA-NeXT MLLM framework for geospatial-temporal sensemaking of remote sensing construction activity.
-
VISOR: A Vision-Language Model-based Test Oracle for Testing Robot
VISOR applies VLMs to automate robot test oracles for correctness and quality assessment while reporting uncertainty, with evaluation on GPT and Gemini showing trade-offs in precision and recall but poor uncertainty calibration.
-
MAG-VLAQ: Multi-modal Aerial-Ground Query Aggregation for Cross-View Place Recognition
MAG-VLAQ fuses multi-modal ground and aerial data via ODE-conditioned vector-of-locally-aggregated-queries to nearly double recall@1 on aerial-ground place recognition benchmarks.
-
Removing the Watermark Is Not Enough: Forensic Stealth in Generative-AI Watermark Removal
Current AI image watermark removal attacks replace the watermark with a different forensic signal, allowing independent detectors to distinguish processed outputs from clean images at over 98% true-positive rate under a 1% false-positive budget.
-
Probability-Flow Distillation: Exact Wasserstein Gradient Flow for High-Fidelity 3D Generation
Probability-Flow Distillation exactly matches the Wasserstein gradient flow of the target distribution when distilling 2D diffusion priors into 3D models, yielding higher-fidelity results than SDS or SDI.
-
Mirror, Mirror on the Wall: Can VLM Agents Tell Who They Are at All?
Stronger VLM agents use mirror reflections for self-identification in controlled 3D tests, while weaker ones inspect but fail to extract or correctly attribute self-relevant information.
-
Object Hallucination-Free Reinforcement Unlearning for Vision-Language Models
HFRU is a two-stage reinforcement unlearning method operating on the vision encoder with GRPO optimization and an abstraction reward that achieves over 98% forgetting and retention on object and face tasks with negligible hallucination.
-
Is Class Signal Clustered or Routed in Task-Induced Implicit Neural Representation Weight Spaces?
Task-induced INR weights are classifiable because their class signal is routed through the reader rather than forming raw geometric clusters.
-
Retrieve, Integrate, and Synthesize: Spatial-Semantic Grounded Latent Visual Reasoning
RIS improves MLLM latent visual reasoning by retrieving spatial-semantic evidence, integrating it via attention bottlenecks, and synthesizing it with language transition tokens, yielding gains on V*, HRBench, MMVP, and BLINK benchmarks.
-
Generalized Category Discovery in Federated Graph Learning
GCD-FGL mitigates neighborhood absorption and global semantic inconsistency in federated generalized category discovery, delivering +4.86 average HRScore gain over baselines on five graph datasets.
-
EAPFusion: Intrinsic Evolving Auxiliary Prior Guidance for Infrared and Visible Image Fusion
EAPFusion uses self-evolving intrinsic priors to produce dynamic, scene-adaptive convolution kernels and channel-mixing fusion for infrared-visible images, reporting state-of-the-art results and downstream gains.