@misc{3dzipspatialawarefeaturediversityguidedtok2026,
  title = {3DZip: Spatial-Aware Feature Diversity-Guided Token Compression for 3D Question Answering},
  year = {2026},
  howpublished = {ECCV 2026},
  url = {https://arxiv.org/abs/2608.01185}
}

@misc{agilepruneranempiricalstudyofattentionandd2026,
  title = {AgilePruner: An Empirical Study of Attention and Diversity for Adaptive Visual Token Pruning in Large Vision-Language Models},
  year = {2026},
  howpublished = {ICLR 2026},
  url = {https://arxiv.org/abs/2603.01236}
}

@misc{avtpmultiimagevisualtokenpruninginlargevis2026,
  title = {AVTP: Multi-Image Visual Token Pruning in Large Visual Language Models},
  year = {2026},
  howpublished = {EMNLP Findings 2026},
  url = {https://arxiv.org/abs/2608.26806}
}

@misc{coverprunerwhospeaksfortheprunedvisualtoke2026,
  title = {CoverPruner: Who Speaks for the Pruned? Visual Token Pruning as Coverage Optimization},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2609.03158}
}

@misc{dstpwhyandwhenvisualtokenpruningfails2026,
  title = {DSTP: Why and When Visual Token Pruning Fails?},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2604.12358}
}

@misc{evocomplearningvisualtokencompressionviase2026,
  title = {EvoComp: Learning Visual Token Compression via Semantic-Guided Evolutionary Labeling},
  year = {2026},
  howpublished = {CVPR 2026},
  url = {https://arxiv.org/abs/2604.17087}
}

@misc{hiloprunelookatwhatyoulllosebeforepruningw2026,
  title = {Hi-Lo Prune: Look at What You'll Lose before Pruning with Hierarchical Token Selection},
  year = {2026},
  howpublished = {CVPR 2026},
  url = {https://openaccess.thecvf.com/content/CVPR2026/html/Sun_Hi-Lo_Prune_Look_at_What_Youll_Lose_before_Pruning_with_CVPR_2026_paper.html}
}

@misc{hybridkvhybridkvcachecompressionforefficie2026,
  title = {HybridKV: Hybrid KV Cache Compression for Efficient Multimodal Large Language Model Inference},
  year = {2026},
  howpublished = {ACL 2026},
  url = {https://arxiv.org/abs/2604.05887}
}

@misc{icctpinstructionguidedcrossmodalclustering2026,
  title = {ICCTP: Instruction-Guided Cross-Modal Clustering for Training-Free Visual Token Pruning},
  year = {2026},
  howpublished = {AAAI 2026},
  url = {https://doi.org/10.1609/aaai.v40i14.38212}
}

@misc{jagdirectboxpredictionforefficientvisualgr2026,
  title = {Jag: Direct Box Prediction for Efficient Visual Grounding},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2610.13239}
}

@misc{mmspecbenchmarkingspeculativedecodingforvi2026,
  title = {MMSpec: Benchmarking Speculative Decoding for Vision-Language Models},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2603.14989}
}

@misc{occamtokenefficientvlminferencewithtrainin2026,
  title = {OccamToken: Efficient VLM Inference with Training-Free and Budget-Adaptive Token Pruning},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2605.29657}
}

@misc{qcachevisualattentionisvaluableinlessthanh2026,
  title = {Q Cache: Visual Attention is Valuable in Less than Half of Decode Layers for Multimodal Large Language Model},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2602.01901}
}

@misc{semanticguidedslowfastpruningofvisualtoken2026,
  title = {Semantic-Guided Slow-Fast Pruning of Visual Tokens for Vision-Language Models},
  year = {2026},
  howpublished = {ICASSP 2026},
  url = {https://ieeexplore.ieee.org/abstract/document/11460400}
}

@misc{sievewhenvisionbecomestext2026,
  title = {SIEVE: When Vision Becomes Text},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2608.10489}
}

@misc{stcacceleratingstreamingvideolargelanguage2026,
  title = {STC: Accelerating Streaming Video Large Language Models via Hierarchical Token Compression},
  year = {2026},
  howpublished = {CVPR 2026},
  url = {https://arxiv.org/abs/2512.00891}
}

@misc{swiftvlmefficientvisionlanguagemodelinfere2026,
  title = {SwiftVLM: Efficient Vision-Language Model Inference via Cross-Layer Token Bypass},
  year = {2026},
  howpublished = {arXiv 2026},
  url = {https://arxiv.org/abs/2602.03134}
}

@misc{unifiedspatiotemporaltokencompressionforvi2026,
  title = {Unified Spatiotemporal Token Compression for Video-LLMs at Ultra-Low Retention},
  year = {2026},
  howpublished = {CVPR 2026},
  url = {https://arxiv.org/abs/2603.21957}
}

@misc{adaptivefastandslowvisualprogramreasoningf2025,
  title = {Adaptive Fast-and-Slow Visual Program Reasoning for Long-Form VideoQA},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2509.17743}
}

@misc{aircacheactivatingintermodalrelevancykvcac2025,
  title = {AirCache: Activating Inter-modal Relevancy KV Cache Compression},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2503.23956}
}

@misc{akvqvlattentionawarekvcacheadaptive2bitqua2025,
  title = {AKVQ-VL: Attention-Aware KV Cache Adaptive 2-Bit Quantization},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2501.15021}
}

@misc{atpllavaadaptivetokenpruningforlvlms2025,
  title = {ATP-LLaVA: Adaptive token pruning for LVLMs},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://openaccess.thecvf.com/content/CVPR2025/html/Ye_ATP-LLaVA_Adaptive_Token_Pruning_for_Large_Vision_Language_Models_CVPR_2025_paper.html}
}

@misc{beyondtextvisualattentionexploitingvisualc2025,
  title = {Beyond Text-Visual Attention: Exploiting Visual Cues for Effective Token Pruning},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2412.01818}
}

@misc{btpbalancedtokenpruning2025,
  title = {BTP: Balanced Token Pruning},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.22038}
}

@misc{calibquant1bitkvcachequantizationformultim2025,
  title = {CalibQuant: 1-Bit KV Cache Quantization for Multimodal LLMs},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2502.14882}
}

@misc{cdprunermaximizingconditionaldiversityfort2025,
  title = {CDPruner: Maximizing Conditional Diversity for Token Pruning},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2506.10967}
}

@misc{dartstoplookingforimportanttokensduplicati2025,
  title = {DART: Stop looking for important tokens, duplication matters more},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2502.11494}
}

@misc{divprunediversitybasedvisualtokenpruning2025,
  title = {DivPrune: Diversity-based visual token pruning},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://openaccess.thecvf.com/content/CVPR2025/html/Alvar_DivPrune_Diversity-based_Visual_Token_Pruning_for_Large_Multimodal_Models_CVPR_2025_paper.html}
}

@misc{dycokedynamiccompressionoftokensforfastvid2025,
  title = {DyCoke: Dynamic Compression of Tokens for Fast Video LLMs},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://openaccess.thecvf.com/content/CVPR2025/html/Tao_DyCoke_Dynamic_Compression_of_Tokens_for_Fast_Video_Large_Language_CVPR_2025_paper.html}
}

@misc{fastcacheoptimizingmultimodalllmserving2025,
  title = {FastCache: Optimizing multimodal LLM serving},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2503.08461}
}

@misc{fastviddynamicdensitypruningforfastvideoll2025,
  title = {FastVID: Dynamic density pruning for fast video LLMs},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2503.11187}
}

@misc{fastvlmefficientvisionencodingforvisionlan2025,
  title = {FastVLM: Efficient vision encoding for vision language models},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://openaccess.thecvf.com/content/CVPR2025/html/Vasu_FastVLM_Efficient_Vision_Encoding_for_Vision_Language_Models_CVPR_2025_paper.html}
}

@misc{fastvlmselfspeculativedecodingforfastvisio2025,
  title = {FastVLM: Self-Speculative Decoding for Fast Vision-Language Model Inference},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.22641}
}

@misc{fitandprunefastandtrainingfreevisualtokenp2025,
  title = {Fit and Prune: Fast and training-free visual token pruning},
  year = {2025},
  howpublished = {AAAI 2025},
  url = {https://doi.org/10.1609/aaai.v39i21.34366}
}

@misc{flashlatentawaresemiautoregressivespeculat2025,
  title = {FLASH: Latent-Aware Semi-Autoregressive Speculative Decoding},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.12728}
}

@misc{focusefficientkeyframeselectionforlongvide2025,
  title = {FOCUS: Efficient Keyframe Selection for Long Video Understanding},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.27280}
}

@misc{folderacceleratingmultimodallargelanguagem2025,
  title = {FOLDER: Accelerating Multi-modal Large Language Models},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2501.02430}
}

@misc{gprunetrainingfreevisualtokenpruningfromgr2025,
  title = {G-Prune: Training-free visual token pruning from graph perspective},
  year = {2025},
  howpublished = {AAAI 2025},
  url = {https://doi.org/10.1609/aaai.v39i4.32427}
}

@misc{glyphscalingcontextwindowsviavisualtextcom2025,
  title = {Glyph: Scaling Context Windows via Visual-Text Compression},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.17800}
}

@misc{hivtphierarchicalvisualtokenpruning2025,
  title = {HIVTP: Hierarchical Visual Token Pruning},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2509.23663}
}

@misc{holitomholistictokenmergingforfastvideollm2025,
  title = {HoliTom: Holistic Token Merging for Fast Video LLMs},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.21334}
}

@misc{infinipotvmemoryconstrainedkvcachecompress2025,
  title = {InfiniPot-V: Memory-Constrained KV Cache Compression},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2506.15745}
}

@misc{livevlmefficientonlinevideounderstandingvi2025,
  title = {LiveVLM: Efficient Online Video Understanding via Streaming-Oriented KV Cache},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.15269}
}

@misc{llavaprumergeadaptivetokenreduction2025,
  title = {LLaVA-PruMerge: Adaptive Token Reduction},
  year = {2025},
  howpublished = {ICCV 2025},
  url = {https://arxiv.org/abs/2403.15388}
}

@misc{longvuspatiotemporaladaptivecompression2025,
  title = {LongVU: Spatiotemporal Adaptive Compression},
  year = {2025},
  howpublished = {ICML 2025},
  url = {https://openreview.net/forum?id=XzZC4gs1mf}
}

@misc{longvuspatiotemporaladaptivecompressionfor2025,
  title = {LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding},
  year = {2025},
  howpublished = {ICML 2025},
  url = {https://openreview.net/forum?id=XzZC4gs1mf}
}

@misc{mllmbasedvideoframeselectionforefficientvi2025,
  title = {M-LLM based video frame selection for efficient video understanding},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://openaccess.thecvf.com/content/CVPR2025/html/Hu_M-LLM_Based_Video_Frame_Selection_for_Efficient_Video_Understanding_CVPR_2025_paper.html}
}

@misc{madakvadaptivemodalityperceptionkvcacheevi2025,
  title = {MadaKV: Adaptive Modality-Perception KV Cache Eviction},
  year = {2025},
  howpublished = {ACL 2025},
  url = {https://aclanthology.org/2025.acl-long.652/}
}

@misc{massvmultimodaladaptationandselfdatadistil2025,
  title = {MASSV: Multimodal Adaptation and Self-Data Distillation for Speculative Decoding of VLMs},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.10526}
}

@misc{medadynamickvcacheallocationforefficientmu2025,
  title = {Meda: Dynamic KV cache allocation for efficient multimodal inference},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2502.17599}
}

@misc{mminferencemodalityawarepermutationsparsea2025,
  title = {MMInference: Modality-Aware Permutation Sparse Attention},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2504.16083}
}

@misc{msdspeculativedecodingreimaginedformultimo2025,
  title = {MSD: Speculative Decoding Reimagined for Multimodal Large Language Models},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.14260}
}

@misc{pixelthinktowardsefficientchainofpixelreas2025,
  title = {PixelThink: Towards Efficient Chain-of-Pixel Reasoning},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.23727}
}

@misc{prolongedreasoningisnotallyouneedcertainty2025,
  title = {Prolonged reasoning is not all you need: Certainty-based adaptive routing},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.15154}
}

@misc{prunevidvisualtokenpruningforefficientvide2025,
  title = {PruneVid: Visual token pruning for efficient video LLMs},
  year = {2025},
  howpublished = {ACL 2025},
  url = {https://aclanthology.org/2025.findings-acl.1024/}
}

@misc{qframequeryawareframeselectionandmultireso2025,
  title = {Q-Frame: Query-aware Frame Selection and Multi-Resolution Adaptation},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2506.22139}
}

@misc{rekvstreamingvideoqawithincontextvideokvca2025,
  title = {ReKV: Streaming video QA with in-context video KV-cache retrieval},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2503.00540}
}

@misc{spargeattnaccuratesparseattentionaccelerat2025,
  title = {SpargeAttn: Accurate sparse attention accelerating any model inference},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2502.18137}
}

@misc{sparsemmheadsparsityemergesfromvisualconce2025,
  title = {SparseMM: Head Sparsity Emerges from Visual Concept Responses},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2506.05344}
}

@misc{sparseviladecouplingvisualsparsityforeffic2025,
  title = {SparseVILA: Decoupling Visual Sparsity for Efficient VLM Inference},
  year = {2025},
  howpublished = {ICCV 2025},
  url = {https://arxiv.org/abs/2510.17777}
}

@misc{specllavaacceleratingvlmswithdynamictreeba2025,
  title = {Spec-LLaVA: Accelerating VLMs with Dynamic Tree-Based Speculative Decoding},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2509.11961}
}

@misc{specvlmenhancingspeculativedecodingviaveri2025,
  title = {SpecVLM: Enhancing speculative decoding via verifier-guided token pruning},
  year = {2025},
  howpublished = {EMNLP 2025},
  url = {https://doi.org/10.48550/arXiv.2508.16201}
}

@misc{specvlmfastspeculativedecodinginvisionlang2025,
  title = {SpecVLM: Fast Speculative Decoding in Vision-Language Models},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2509.11815}
}

@misc{st3acceleratingmllmbyspatialtemporalvisual2025,
  title = {ST3: Accelerating MLLM by spatial-temporal visual token trimming},
  year = {2025},
  howpublished = {AAAI 2025},
  url = {https://doi.org/10.1609/aaai.v39i10.33201}
}

@misc{streamingtomstreamingtokencompression2025,
  title = {StreamingTOM: Streaming Token Compression},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.18269}
}

@misc{streamingvlmrealtimeunderstandingforinfini2025,
  title = {StreamingVLM: Real-time understanding for infinite video streams},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.09608}
}

@misc{sttmmultigranularspatiotemporaltokenmergin2025,
  title = {STTM: Multi-granular spatio-temporal token merging},
  year = {2025},
  howpublished = {ICCV 2025},
  url = {https://doi.org/10.48550/arXiv.2507.07990}
}

@misc{timechatonline80visualtokensarenaturallyre2025,
  title = {TimeChat-Online: 80\% Visual Tokens are Naturally Redundant},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://doi.org/10.48550/ARXIV.2504.17343}
}

@misc{tokenpackerefficientvisualprojectorformult2025,
  title = {TokenPacker: Efficient Visual Projector for Multimodal LLM},
  year = {2025},
  howpublished = {IJCV 2025},
  url = {https://doi.org/10.1007/S11263-025-02491-7}
}

@misc{tosatokenmergingwithspatialawareness2025,
  title = {ToSA: Token Merging with Spatial Awareness},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2506.20066}
}

@misc{vicoatrainingstrategytowardssemanticawared2025,
  title = {ViCO: A Training Strategy towards Semantic Aware Dynamic High-Resolution},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.12793}
}

@misc{vidcom2videocompressioncommander2025,
  title = {VidCom2: Video Compression Commander},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2505.14454}
}

@misc{videoxl2towardsverylongvideounderstandingt2025,
  title = {Video-XL-2: Towards Very Long-Video Understanding Through Task-Aware KV Sparsification},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2506.19225}
}

@misc{videonsanativesparseattentionscalesvideoun2025,
  title = {VideoNSA: Native Sparse Attention Scales Video Understanding},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2510.02295}
}

@misc{videotreeadaptivetreebasedvideorepresentat2025,
  title = {VideoTree: Adaptive tree-based video representation},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://openaccess.thecvf.com/content/CVPR2025/html/Wang_VideoTree_Adaptive_Tree-based_Video_Representation_for_LLM_Reasoning_on_Long_CVPR_2025_paper.html}
}

@misc{vidkvplugandplay1xbitkvcachequantization2025,
  title = {VidKV: Plug-and-Play 1.x-Bit KV Cache Quantization},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2503.16257}
}

@misc{visionthinksmartandefficientvisionlanguage2025,
  title = {VisionThink: Smart and efficient vision language model via reinforcement learning},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://doi.org/10.48550/arXiv.2507.13348}
}

@misc{visionziplongerisbetterbutnotnecessary2025,
  title = {VisionZip: Longer is Better but Not Necessary},
  year = {2025},
  howpublished = {CVPR 2025},
  url = {https://arxiv.org/abs/2412.04467}
}

@misc{vispecacceleratingvlmswithvisionawarespecu2025,
  title = {ViSpec: Accelerating VLMs with Vision-Aware Speculative Decoding},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2509.15235}
}

@misc{xattentionblocksparseattentionwithantidiag2025,
  title = {XAttention: Block sparse attention with antidiagonal scoring},
  year = {2025},
  howpublished = {arXiv 2025},
  url = {https://arxiv.org/abs/2503.16428}
}

@misc{aimadaptiveinferenceofmultimodalllms2024,
  title = {AIM: Adaptive Inference of Multi-Modal LLMs},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://doi.org/10.48550/ARXIV.2412.03248}
}

@misc{convllavahierarchicalbackbonesasvisualenco2024,
  title = {ConvLLaVA: Hierarchical backbones as visual encoder for large multimodal models},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2405.15738}
}

@misc{dynamicvlmsimpledynamicvisualtokencompress2024,
  title = {Dynamic-VLM: Simple Dynamic Visual Token Compression},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://doi.org/10.48550/ARXIV.2412.09530}
}

@misc{dynamicvlmsimpledynamicvisualtokencompress2024,
  title = {Dynamic-VLM: Simple Dynamic Visual Token Compression for VideoLLM},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://doi.org/10.48550/ARXIV.2412.09530}
}

@misc{elasticcacheefficientinferenceofvisioninst2024,
  title = {Elastic Cache: Efficient inference of vision instruction-following models},
  year = {2024},
  howpublished = {ECCV 2024},
  url = {https://doi.org/10.1007/978-3-031-72643-9_4}
}

@misc{fastvanimageisworth12tokensafterlayer22024,
  title = {FastV: An image is worth 1/2 tokens after layer 2},
  year = {2024},
  howpublished = {ECCV 2024},
  url = {https://doi.org/10.1007/978-3-031-73004-7_2}
}

@misc{framevoyagerlearningtoqueryframesforvideol2024,
  title = {Frame-Voyager: Learning to query frames for video LLMs},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2410.03226}
}

@misc{framefusioncombiningsimilarityandimportanc2024,
  title = {FrameFusion: Combining similarity and importance},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2501.01986}
}

@misc{keyvideollmtowardslargescalevideokeyframes2024,
  title = {KeyVideoLLM: Towards large-scale video keyframe selection},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2407.03104}
}

@misc{lessismoreasimpleyeteffectivetokenreductio2024,
  title = {Less is More: A Simple yet Effective Token Reduction Method},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2409.10994}
}

@misc{longcontexttransferfromlanguagetovision2024,
  title = {Long context transfer from language to vision},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2406.16852}
}

@misc{longvlmefficientlongvideounderstandingvial2024,
  title = {LongVLM: Efficient long video understanding via large language models},
  year = {2024},
  howpublished = {ECCV 2024},
  url = {https://doi.org/10.1007/978-3-031-73414-4_26}
}

@misc{lookmlookonceoptimizationinkvcache2024,
  title = {LOOK-M: Look-once optimization in KV cache},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2406.18139}
}

@misc{moviechatfromdensetokentosparsememoryforlo2024,
  title = {MovieChat: From dense token to sparse memory for long video understanding},
  year = {2024},
  howpublished = {CVPR 2024},
  url = {https://doi.org/10.1109/CVPR52733.2024.01725}
}

@misc{pvcprogressivevisualtokencompression2024,
  title = {PVC: Progressive Visual Token Compression},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2412.09613}
}

@misc{pyramiddropacceleratingviapyramidvisualred2024,
  title = {PyramidDrop: Accelerating via pyramid visual redundancy reduction},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2410.17247}
}

@misc{qwen2vlenhancingvisionlanguagemodelspercep2024,
  title = {Qwen2-VL: Enhancing vision-language model's perception of the world at any resolution},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2409.12191}
}

@misc{sparsevlmvisualtokensparsificationforeffic2024,
  title = {SparseVLM: Visual token sparsification for efficient inference},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2410.04417}
}

@misc{videochatgpttowardsdetailedvideounderstand2024,
  title = {Video-ChatGPT: Towards detailed video understanding via large vision and language models},
  year = {2024},
  howpublished = {ACL 2024},
  url = {https://doi.org/10.18653/v1/2024.acl-long.679}
}

@misc{vilaefficientvideolanguagealignmentforvide2024,
  title = {VILA: Efficient video-language alignment for video question answering},
  year = {2024},
  howpublished = {ECCV 2024},
  url = {https://doi.org/10.1007/978-3-031-73033-7_11}
}

@misc{vlcachesparsityandmodalityawarekvcachecomp2024,
  title = {VL-Cache: Sparsity and modality-aware KV cache compression},
  year = {2024},
  howpublished = {arXiv 2024},
  url = {https://arxiv.org/abs/2410.23317}
}

@misc{blip2bootstrappinglanguageimagepretraining2023,
  title = {BLIP-2: Bootstrapping Language-Image Pre-training},
  year = {2023},
  howpublished = {ICML 2023},
  url = {https://proceedings.mlr.press/v202/li23q.html}
}

@misc{fastvitafasthybridvisiontransformerusingst2023,
  title = {FastViT: A fast hybrid vision transformer using structural reparameterization},
  year = {2023},
  howpublished = {ICCV 2023},
  url = {https://doi.org/10.1109/ICCV51070.2023.00532}
}

@misc{sevilaselfchainedimagelanguagemodelforvide2023,
  title = {SeViLA: Self-chained image-language model for video localization},
  year = {2023},
  howpublished = {NeurIPS 2023},
  url = {http://papers.nips.cc/paper_files/paper/2023/hash/f22a9af8dbb348952b08bd58d4734b50-Abstract-Conference.html}
}

@misc{videollamaaninstructiontunedaudiovisuallan2023,
  title = {Video-LLaMA: An instruction-tuned audio-visual language model},
  year = {2023},
  howpublished = {arXiv 2023},
  url = {https://arxiv.org/abs/2306.02858}
}