mlsanityData to 14 Oct 2026

Temporal Preference Optimization for Long-Form Video Understanding

arXiv 2501.13919 ↗First posted 23 Jan 2025cs.CV cs.AI cs.CLcited by 29

On the map: Multimodal LLMs / vision-language models → Video understanding with MLLMs

also in 1 more
Abstract (arXiv)

Despite significant advancements in video large multimodal models (video-LMMs), achieving effective temporal grounding in long-form videos remains a challenge for existing models. To address this limitation, we propose Temporal Preference Optimization (TPO), a novel post-training framework designed to enhance the temporal grounding capabilities of video-LMMs through preference learning. TPO adopts a self-training approach that enables models to differentiate between well-grounded and less accurate temporal responses by leveraging curated preference datasets at two granularities: localized temporal grounding, which focuses on specific video segments, and comprehensive temporal grounding, which captures extended temporal dependencies across entire video sequences. By optimizing on these preference datasets, TPO significantly enhances temporal understanding while reducing reliance on manually annotated data. Extensive experiments on three long-form video understanding benchmarks--LongVideoBench, MLVU, and Video-MME--demonstrate the effectiveness of TPO across two state-of-the-art video-LMMs. Notably, LLaVA-Video-TPO establishes itself as the leading 7B model on the Video-MME benchmark, underscoring the potential of TPO as a scalable and efficient solution for advancing temporal reasoning in long-form video understanding. Project page: https://ruili33.github.io/tpo_website.

Key facts. Cited by 29 papers in the map (data through 14 Oct 2026). Placed in Video understanding with MLLMs, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 58 of its 69 references that are in the map. First posted on arXiv 23 Jan 2025.

Builds on

58 of 69 references.

Closest ancestors · 14

Same direction, newest first

  1. 2024Apollo: An Exploration of Video Understanding in Large Multimodal Models13 Dec
  2. 2024TimeMarker: A Versatile Video-LLM for Long and Short Video Understanding with Superior Temporal Localization Ability27 Nov
  3. 2024LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding22 Oct
  4. 2024Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understanding22 Sep
  5. 2024Kangaroo: A Powerful Video-Language Model Supporting Long-context Video Input28 Aug
  6. 2024Video-CCAM: Enhancing Video-Language Understanding with Causal Cross-Attention Masks for Short and Long Videos26 Aug
  7. 2024LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding22 Jul
  8. 2024Video-STaR: Self-Training Enables Video Instruction Tuning with Any Supervision8 Jul
  9. 2024Long Context Transfer from Language to Vision24 Jun
  10. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis31 May
  11. 2024LITA: Language Instructed Temporal-Localization Assistant27 Mar
  12. 2023TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding4 Dec
2 more
  1. 2023Video-LLaVA: Learning United Visual Representation by Alignment Before Projection16 Nov
  2. 2023Vid2Seq: Large-Scale Pretraining of a Visual Language Model for Dense Video Captioning27 Feb

Other references · 44

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  4. 2023Visual Instruction Tuningcited by 7,869
  5. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  6. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  7. 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,568
  8. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
  9. 2024LLaVA-OneVision: Easy Visual Task Transfercited by 2,719
  10. 2020Learning to summarize from human feedbackcited by 2,588
  11. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Taskscited by 2,565
  12. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,224
32 more
  1. 2019Fine-Tuning Language Models from Human Preferencescited by 2,150
  2. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,917
  3. 2024MiniCPM-V: A GPT-4V Level MLLM on Your Phonecited by 1,079
  4. 2022STaR: Bootstrapping Reasoning With Reasoningcited by 940
  5. 2023VideoChat: Chat-Centric Video Understandingcited by 857
  6. 2023Aligning Large Multimodal Models with Factually Augmented RLHFcited by 712
  7. 2023VILA: On Pre-training for Visual Language Modelscited by 692
  8. 2022Large Language Models Can Self-Improvecited by 664
  9. 2017TALL: Temporal Activity Localization via Language Querycited by 584
  10. 2024ShareGPT4Video: Improving Video Understanding and Generation with Better Captionscited by 388
  11. 2024LMMs-Eval: Reality Check on the Evaluation of Large Multimodal Modelscited by 353
  12. 2024VideoAgent: Long-form Video Understanding with Large Language Model as Agentcited by 332
  13. 2022Large Language Models Are Reasoning Teacherscited by 329
  14. 2023Detecting and Preventing Hallucinations in Large Vision Language Modelscited by 256
  15. 2023Beyond Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimizationcited by 225
  16. 2024CogVLM2: Visual Language Models for Image and Video Understandingcited by 217
  17. 2024Aligning Modalities in Vision Large Language Models via Preference Fine-tuningcited by 198
  18. 2025VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interactioncited by 197
  19. 2023Query-Dependent Video Representation for Moment Retrieval and Highlight Detectioncited by 166
  20. 2024Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolutioncited by 159
  21. 2024Direct Preference Optimization of Video Large Multimodal Models from Language Model Rewardcited by 147
  22. 2024Aria: An Open Multimodal Native Mixture-of-Experts Modelcited by 138
  23. 2019Semantic Conditioned Dynamic Modulation for Temporal Sentence Grounding in Videoscited by 135
  24. 2023Can I Trust Your Answer? Visually Grounded Video Question Answeringcited by 135
  25. 2021End-to-End Dense Video Captioning with Parallel Decodingcited by 133
  26. 2023Silkie: Preference Distillation for Large Visual Language Modelscited by 130
  27. 2015Every Moment Counts: Dense Detailed Labeling of Actions in Complex Videoscited by 93
  28. 2024Grounded-VideoLLM: Sharpening Fine-grained Temporal Grounding in Video Large Language Modelscited by 91
  29. 2024Strengthening Multimodal Large Language Model with Bootstrapped Preference Optimizationcited by 75
  30. 2024Enhancing Large Vision Language Models with Self-Training on Image Comprehensioncited by 70
  31. 2024Tuning Large Multimodal Models for Videos using Reinforcement Learning from AI Feedbackcited by 36
  32. 2023Understanding Transferable Representation Learning and Zero-shot Transfer in CLIPcited by 26

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 22; 2026: 7.

29 papers.

Sort:

By area:

  1. 2025SmolVLM: Redefining small and efficient multimodal modelscited by 237
  2. 2025Unified Reward Model for Multimodal Understanding and Generationcited by 199
  3. 2025TSPO: Temporal Sampling Policy Optimization for Long-form Video Language Understandingcited by 25
  4. 2025MR. Video: "MapReduce" is the Principle for Long Video Understandingcited by 23
  5. 2025FlexSelect: Flexible Token Selection for Efficient Long Video Understandingcited by 18
  6. 2025Hallo4: High-Fidelity Dynamic Portrait Animation via Direct Preference Optimizationcited by 16
  7. 2025TEMPURA: Temporal Event Masked Prediction and Understanding for Reasoning in Actioncited by 14
  8. 2025Agentic Keyframe Search for Video Question Answeringcited by 13
  9. 2025AVATAR: Reinforcement Learning to See, Hear, and Reason Over Videocited by 12
  10. 2025SEASON: Mitigating Temporal Hallucination in Video Large Language Models via Self-Diagnostic Contrastive Decodingcited by 10
  11. 2025Tool-Augmented Spatiotemporal Reasoning for Streamlining Video Question Answering Taskcited by 10
  12. 2025VideoPASTA: 7K Preference Pairs That Matter for Video-LLM Alignmentcited by 8
17 more
  1. 2025Zoom-Zero: Reinforced Coarse-to-Fine Video Understanding via Temporal Zoom-incited by 8
  2. 2025Self-alignment of Large Video Language Models with Refined Regularized Preference Optimizationcited by 7
  3. 2025VideoVista-CulturalLingo: 360^circ Horizons-Bridging Cultures, Languages, and Domains in Video Comprehensioncited by 4
  4. 2025Factorized Learning for Temporally Grounded Video-Language Modelscited by 4
  5. 2025TEMPLE: Incentivizing Temporal Understanding of Video Large Language Models via Progressive Pre-SFT Alignmentcited by 3
  6. 2025SynPO: Synergizing Descriptiveness and Preference Optimization for Video Detailed Captioningcited by 3
  7. 2026AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video Understandingcited by 2
  8. 2026Reinforcing Consistency in Video MLLMs with Structured Rewardscited by 2
  9. 2025LeanPO: Lean Preference Optimization for Likelihood Alignment in Video-LLMscited by 1
  10. 2025How Important are Videos for Training Video LLMs?cited by 1
  11. 2025Benchmarking the Trustworthiness in Multimodal LLMs for Video Understandingcited by 1
  12. 2026LongVPO: From Anchored Cues to Self-Reasoning for Long-Form Video Preference Optimizationcited by 1
  13. 2026Towards Temporal Compositional Reasoning in Long-Form Sports Videoscited by 1
  14. 2025Oracle-RLAIF: An Improved Fine-Tuning Framework for Multi-modal Video Models using Reinforcement Learning from Ranking Feedbackcited by 0
  15. 2026Bridging VideoQA and Video-Guided Agentic Tasks via Generalized Keyframe Extractioncited by 0
  16. 2026VADER: Adaptive Debiasing for Hallucination Mitigation in Video Large Language Modelscited by 0
  17. 2026Multi-Agent Self-Improving Reinforcement Learning for Video Reasoningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Harnessing Synthetic Preference Data for Enhancing Temporal Understanding of Video-LLMs
  2. 2025Tempo-R0: A Video-MLLM for Temporal Video Grounding through Efficient Temporal Sensing Reinforcement Learning
  3. 2025DenseDPO: Fine-Grained Temporal Preference Optimization for Video Diffusion Models
  4. 2025TimeLens: Rethinking Video Temporal Grounding with Multimodal LLMs
  5. 2025VistaDPO: Video Hierarchical Spatial-Temporal Direct Preference Optimization for Large Video Models
  6. 2025A Survey on Video Temporal Grounding with Multimodal Large Language Model
  7. 2025VideoITG: Multimodal Video Understanding with Instructed Temporal Grounding
  8. 2025Investigating and Enhancing the Robustness of Large Multimodal Models Against Temporal Inconsistency