SYNCR benchmark shows leading MLLMs reach only 52.5% average accuracy on cross-video reasoning tasks against an 89.5% human baseline, with major weaknesses in physical and spatial reasoning.
hub
Pllava: Parameter-free llava extension from images to videos for video dense captioning
17 Pith papers cite this work. Polarity classification is still indexing.
17
Pith papers citing it
hub tools
citation-role summary
background 1
baseline 1
citation-polarity summary
verdicts
UNVERDICTED 17representative citing papers
Sink-Token-aware Pruning (SToP) suppresses semantically uninformative sink tokens during visual token pruning in Video LLMs, boosting fine-grained performance even at 90% pruning rates across hallucination, reasoning, and MCQA benchmarks.
Gaze Attention groups visual embeddings into selectable regions and dynamically restricts attention to task-relevant ones, matching dense baselines with up to 90% fewer visual KV entries via added context tokens.
VideoRouter uses dual semantic and image routers for query-adaptive token compression in long-video models, delivering up to 67.9% reduction while outperforming the InternVL baseline on VideoMME, MLVU, and LongVideoBench.
WindowQuant performs window-adaptive mixed-precision KV cache quantization guided by similarity to the text prompt, with reordering to enable efficient inference in VLMs.
XComp reaches extreme video compression (one token per selective frame) via learnable progressive token compression and question-conditioned frame selection, lifting LVBench accuracy from 42.9 percent to 46.2 percent after tuning on 2.5 percent of standard data.
HTDC mitigates hallucinations in LVLMs by triggering calibration only at hesitation-prone decoding steps via contrasts with visual-nullification and semantic-nullification probes.
Tempo uses a 6B SVLM as a local temporal compressor with training-free adaptive token allocation to achieve SOTA long-video understanding at 0.5-16 tokens per frame, scoring 52.3 on 4101s LVBench under 8K budget.
STEAR reduces spatial and temporal hallucinations in Video-LLMs via layer-aware evidence intervention from middle decoder layers in a single-encode pass.
ProVCA progressively condenses long videos via segment localization, snippet selection, and keyframe refinement to achieve SOTA zero-shot accuracies on EgoSchema, NExT-QA, and IntentQA with fewer frames.
LLaVA-Video-178K is a new synthetic video instruction dataset that, when combined with existing data to train LLaVA-Video, produces strong results on video understanding benchmarks.
Open-Sora releases an open-source video generation model based on a Spatial-Temporal Diffusion Transformer that decouples spatial and temporal attention, supporting text-to-video, image-to-video, and text-to-image tasks with claimed high fidelity.
LLaVA-OneVision is the first single open LMM to simultaneously achieve strong performance in single-image, multi-image, and video scenarios with cross-scenario transfer capabilities.
The paper offers the first focused review of MLLM-based video translation organized by a three-role taxonomy of Semantic Reasoner, Expressive Performer, and Visual Synthesizer, plus open challenges.
Show-o2 unifies text, image, and video understanding and generation in a single autoregressive-plus-flow-matching model built on 3D causal VAE representations.
VideoLLaMA3 uses a vision-centric training paradigm and token-reduction design to reach competitive results on image and video benchmarks.
VideoLLaMA 2 improves video LLMs via a new STC connector for spatial-temporal dynamics and joint audio training, reaching competitive results on video QA and captioning benchmarks.
citing papers explorer
-
SYNCR: A Cross-Video Reasoning Benchmark with Synthetic Grounding
SYNCR benchmark shows leading MLLMs reach only 52.5% average accuracy on cross-video reasoning tasks against an 89.5% human baseline, with major weaknesses in physical and spatial reasoning.
-
Sink-Token-Aware Pruning for Fine-Grained Video Understanding in Efficient Video LLMs
Sink-Token-aware Pruning (SToP) suppresses semantically uninformative sink tokens during visual token pruning in Video LLMs, boosting fine-grained performance even at 90% pruning rates across hallucination, reasoning, and MCQA benchmarks.
-
Learning to See What You Need: Gaze Attention for Multimodal Large Language Models
Gaze Attention groups visual embeddings into selectable regions and dynamically restricts attention to task-relevant ones, matching dense baselines with up to 90% fewer visual KV entries via added context tokens.
-
VideoRouter: Query-Adaptive Dual Routing for Efficient Long-Video Understanding
VideoRouter uses dual semantic and image routers for query-adaptive token compression in long-video models, delivering up to 67.9% reduction while outperforming the InternVL baseline on VideoMME, MLVU, and LongVideoBench.
-
WindowQuant: Mixed-Precision KV Cache Quantization based on Window-Level Similarity for VLMs Inference Optimization
WindowQuant performs window-adaptive mixed-precision KV cache quantization guided by similarity to the text prompt, with reordering to enable efficient inference in VLMs.
-
One Token per Highly Selective Frame: Towards Extreme Compression for Long Video Understanding
XComp reaches extreme video compression (one token per selective frame) via learnable progressive token compression and question-conditioned frame selection, lifting LVBench accuracy from 42.9 percent to 46.2 percent after tuning on 2.5 percent of standard data.
-
HTDC: Hesitation-Triggered Differential Calibration for Mitigating Hallucination in Large Vision-Language Models
HTDC mitigates hallucinations in LVLMs by triggering calibration only at hesitation-prone decoding steps via contrasts with visual-nullification and semantic-nullification probes.
-
Small Vision-Language Models are Smart Compressors for Long Video Understanding
Tempo uses a 6B SVLM as a local temporal compressor with training-free adaptive token allocation to achieve SOTA long-video understanding at 0.5-16 tokens per frame, scoring 52.3 on 4101s LVBench under 8K budget.
-
STEAR: Layer-Aware Spatiotemporal Evidence Intervention for Hallucination Mitigation in Video Large Language Models
STEAR reduces spatial and temporal hallucinations in Video-LLMs via layer-aware evidence intervention from middle decoder layers in a single-encode pass.
-
Progressive Video Condensation with MLLM Agent for Long-form Video Understanding
ProVCA progressively condenses long videos via segment localization, snippet selection, and keyframe refinement to achieve SOTA zero-shot accuracies on EgoSchema, NExT-QA, and IntentQA with fewer frames.
-
Empowering Video Translation using Multimodal Large Language Models
The paper offers the first focused review of MLLM-based video translation organized by a three-role taxonomy of Semantic Reasoner, Expressive Performer, and Visual Synthesizer, plus open challenges.