mlsanityData to 7 Oct 2026

End-to-End Training for Autoregressive Video Diffusion via Self-Resampling

arXiv 2512.15702 ↗First posted 17 Dec 2025cs.CVcited by 45

On the map: Diffusion models → Video generation

Abstract (arXiv)

Autoregressive video diffusion models hold promise for world simulation but are vulnerable to exposure bias arising from the train-test mismatch. While recent works address this via post-training, they typically rely on a bidirectional teacher model or discriminator. To achieve an end-to-end solution, we introduce Resampling Forcing, a teacher-free framework that enables training autoregressive video models from scratch and at scale. Central to our approach is a self-resampling scheme that simulates inference-time model errors on history frames during training. Conditioned on these degraded histories, a sparse causal mask enforces temporal causality while enabling parallel training with frame-level diffusion loss. To facilitate efficient long-horizon generation, we further introduce history routing, a parameter-free mechanism that dynamically retrieves the top-k most relevant history frames for each query. Experiments demonstrate that our approach achieves performance comparable to distillation-based baselines while exhibiting superior temporal consistency on longer videos owing to native-length training.

Key facts. Cited by 45 papers in the map (data through 7 Oct 2026). Placed in Video generation, one of 14 directions in Diffusion models. Builds on 79 of its 83 references that are in the map. First posted on arXiv 17 Dec 2025.

Builds on

79 of 83 references.

Closest ancestors · 35

Same direction, newest first

  1. 2025Stable Video Infinity: Infinite-Length Video Generation with Error Recycling10 Oct
  2. 2025LinVideo: A Post-Training Framework towards O(n) Attention in Efficient Video Generation9 Oct
  3. 2025Self-Forcing++: Towards Minute-Scale High-Quality Video Generation2 Oct
  4. 2025SANA-Video: Efficient Video Generation with Block Linear Diffusion Transformer29 Sep
  5. 2025VMoBA: Mixture-of-Block Attention for Video Diffusion Models30 Jun
  6. 2025Radial Attention: O(nlog n) Sparse Attention with Energy Decay for Long Video Generation24 Jun
  7. 2025Seedance 1.0: Exploring the Boundaries of Video Generation Models10 Jun
  8. 2025Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion9 Jun
  9. 2025Test-Time Training Done Right29 May
  10. 2025Long-Context State-Space Video World Models26 May
  11. 2025VORTA: Efficient Video Diffusion via Routing Sparse Attention24 May
  12. 2025VSA: Faster Video Diffusion with Trainable Sparse Attention19 May
23 more
  1. 2025Generative Pre-trained Autoregressive Diffusion Transformer12 May
  2. 2025SkyReels-V2: Infinite-length Film Generative Model17 Apr
  3. 2025Long Context Tuning for Video Generation13 Mar
  4. 2025Error Analyses of Auto-Regressive Video Diffusion Models: A Unified Framework12 Mar
  5. 2025AR-Diffusion: Asynchronous Video Generation with Auto-Regressive Diffusion10 Mar
  6. 2025Training-free and Adaptive Sparse Attention for Efficient Long Video Generation28 Feb
  7. 2025Sparse VideoGen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity3 Feb
  8. 2024LTX-Video: Realtime Video Latent Diffusion30 Dec
  9. 2024Autoregressive Video Generation without Vector Quantization18 Dec
  10. 2024LinGen: Towards High-Resolution Minute-Length Text-to-Video Generation with Linear Computational Complexity13 Dec
  11. 2024Ca2-VDM: Efficient Autoregressive Video Diffusion Model with Causal Generation and Cache Sharing25 Nov
  12. 2024ARLON: Boosting Diffusion Transformers with Autoregressive Models for Long Video Generation27 Oct
  13. 2024MarDini: Masked Autoregressive Diffusion for Video Generation at Scale26 Oct
  14. 2024Progressive Autoregressive Video Diffusion Models10 Oct
  15. 2024Pyramidal Flow Matching for Efficient Video Generative Modeling8 Oct
  16. 2024Diffusion Models Are Real-Time Game Engines27 Aug
  17. 2024CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer12 Aug
  18. 2024StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text21 Mar
  19. 2024Lumiere: A Space-Time Diffusion Model for Video Generation23 Jan
  20. 2023ARTboldsymbol·V: Auto-Regressive Text-to-Video Generation with Diffusion Models30 Nov
  21. 2023Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets25 Nov
  22. 2023AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning10 Jul
  23. 2023Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models18 Apr

Other references · 44

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  4. 2022Scalable Diffusion Models with Transformerscited by 5,931
  5. 2022Generative Adversarial Networkscited by 5,603
  6. 2022Flow Matching for Generative Modelingcited by 5,426
  7. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,094
  8. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,611
  9. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,551
  10. 2025Wan: Open and Advanced Large-Scale Video Generative Modelscited by 2,536
  11. 2023VBench: Comprehensive Benchmark Suite for Video Generative Modelscited by 1,755
  12. 2024HunyuanVideo: A Systematic Framework For Large Video Generative Modelscited by 1,597
32 more
  1. 2023One-step Diffusion with Distribution Matching Distillationcited by 978
  2. 2015Scheduled Sampling for Sequence Prediction with Recurrent Neural Networkscited by 779
  3. 2024Improved Distribution Matching Distillation for Fast Image Synthesiscited by 648
  4. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 634
  5. 2017MoCoGAN: Decomposing Motion and Content for Video Generationcited by 523
  6. 2025Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attentioncited by 391
  7. 2022Analog Bits: Generating Discrete Data using Diffusion Models with Self-Conditioningcited by 362
  8. 2021StyleGAN-V: A Continuous Video Generator with the Price, Image Quality and Perks of StyleGAN2cited by 300
  9. 2024Diffusion for World Modeling: Visual Details Matter in Ataricited by 285
  10. 2025MAGI-1: Autoregressive Video Generation at Scalecited by 262
  11. 2025LongLive: Real-time Interactive Long Video Generationcited by 237
  12. 2024Score identity Distillation: Exponentially Fast Distillation of Pretrained Diffusion Models for One-Step Generationcited by 211
  13. 2025Video models are zero-shot learners and reasonerscited by 208
  14. 2024How Far is Video Generation from World Model: A Physical Law Perspectivecited by 204
  15. 2025MoBA: Mixture of Block Attention for Long-Context LLMscited by 188
  16. 2025History-Guided Video Diffusioncited by 165
  17. 2025GameFactory: Creating New Games with Generative Interactive Videoscited by 124
  18. 2019Generalization in Generation: A closer look at Exposure Biascited by 123
  19. 2024Rolling Diffusion Modelscited by 102
  20. 2023Elucidating the Exposure Bias in Diffusion Modelscited by 98
  21. 2025Mixture of Contexts for Long Video Generationcited by 90
  22. 2025Autoregressive Adversarial Post-Training for Real-Time Interactive Video Generationcited by 84
  23. 2025MotionStream: Real-Time Video Generation with Interactive Motion Controlscited by 71
  24. 2025A Fourier Space Perspective on Diffusion Modelscited by 48
  25. 2024Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Stepcited by 47
  26. 2019Scheduled Sampling for Transformerscited by 46
  27. 2024ACDiT: Interpolating Autoregressive Conditional Modeling and Diffusion Transformercited by 45
  28. 2025Pre-Trained Video Generative Models as World Simulatorscited by 43
  29. 2025InfinityStar: Unified Spacetime AutoRegressive Modeling for Visual Generationcited by 43
  30. 2025STARFlow-V: End-to-End Video Generative Modeling with Normalizing Flowscited by 9
  31. 2024Bridging the Training-Inference Gap in LLMs by Leveraging Self-Generated Tokenscited by 7
  32. 2025Autoregressive Video Generation beyond Next Frames Predictioncited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 45.

45 papers.

Sort:

By area:

  1. 2026Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generationcited by 159
  2. 2026Helios: Real Real-Time Long Video Generation Modelcited by 56
  3. 2026Causal Forcing++: Scalable Few-Step Autoregressive Diffusion Distillation for Real-Time Interactive Video Generationcited by 44
  4. 2026LoL: Longer than Longer, Scaling Video Generation to Hourcited by 29
  5. 2026AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillationcited by 23
  6. 2026Relax Forcing: Relaxed KV-Memory for Consistent Long Video Generationcited by 20
  7. 2026LIVE: Long-horizon Interactive Video World Modelingcited by 14
  8. 2026ActWorld: From Explorable to Interactive World Model via Action-Aware Memorycited by 13
  9. 2026Pathwise Test-Time Correction for Autoregressive Long Video Generationcited by 12
  10. 2026Mode Seeking meets Mean Seeking for Fast Long Video Generationcited by 12
  11. 2026CausalCine: Real-Time Autoregressive Generation for Multi-Shot Video Narrativescited by 8
  12. 2026A Mechanistic View on Video Generation as World Models: State and Dynamicscited by 6
33 more
  1. 2026Efficient Video Diffusion Models: Advancements and Challengescited by 6
  2. 2026RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPOcited by 6
  3. 2026One Sentence, One Drama: Personalized Short-Form Drama Generation via Multi-Agent Systemscited by 6
  4. 2026Grounded Forcing: Bridging Time-Independent Semantics and Proximal Dynamics in Autoregressive Video Synthesiscited by 5
  5. 2026Head Forcing: Long Autoregressive Video Generation via Head Heterogeneitycited by 5
  6. 2026DySink: Dynamic Frame Sinks for Autoregressive Long Video Generationcited by 5
  7. 2026Context-Matched Distillation: Teacher Causality for Autoregressive Video Distillationcited by 5
  8. 2026Ring Forcing: Towards Precise Long-Term Memory for Autoregressive Video Diffusioncited by 5
  9. 2026MaineCoon: Pursuing A Real-Time Audio-Visual Social World Modelcited by 4
  10. 2026RoboWorld: Fast and Reliable Neural Simulators for Generalist Robot Policy Evaluationcited by 4
  11. 2026From Pixels to States: Rethinking Interactive World Models as Game Enginescited by 4
  12. 2026JoyAI-Video-Edit: Real-Time Open-Ended Video Editing with Autoregressive Diffusioncited by 4
  13. 2026FAR-Drive: Frame-AutoRegressive Video Generation in Closed-Loop Autonomous Drivingcited by 3
  14. 2026HorizonDrive: Self-Corrective Autoregressive World Model for Long-horizon Driving Simulationcited by 3
  15. 2026SoulX-LiveAct: Towards Hour-Scale Real-Time Human Animation with Neighbor Forcing and ConvKV Memorycited by 2
  16. 2026GRNEdit: Efficient General Video Editing from a New Binary-Evidence Perspective in Generative Refinement Networkscited by 2
  17. 2026TokenTrim: Inference-Time Token Pruning for Autoregressive Long Video Generationcited by 1
  18. 2026Precise Editing and Flexible Referencing for Interactable Worldscited by 1
  19. 2026Robust Dreamer: Deviation-Aware Latent Gaussian Memory for Action-Controlled AR Video Generationcited by 0
  20. 2026LeapTalk: Breaking the Latency-Quality Trade-off in Talking Head Generationcited by 0
  21. 2026WALL-SS: Scaling Long-horizon World Models via Next-Scale Autoregressioncited by 0
  22. 2026Recency Forcing: Bridging the Long-Horizon Gap in Autoregressive Video Generationcited by 0
  23. 2026Rethinking Streaming Video Diffusion Model: Context, Execution, and Trainingcited by 0
  24. 2026SewFusion: Tailored Generation of Topology and Panel-Level Geometry for Sewing Patternscited by 0
  25. 2026OneFixer: High-Quality and Consistent One-Step Autoregressive 3DGS Refinement for Driving Scenescited by 0
  26. 2026Triangular Resampling for Long-Horizon Motion Generationcited by 0
  27. 2026From Scores to Samples: Elastic Forcing for Autoregressive Video Generationcited by 0
  28. 2026ReCaVSR: One-Step Streaming Diffusion Video Super-Resolution with Recycled Latents and Learned Cache Routingcited by 0
  29. 2026Self-Aligned Forcing: Streaming Video Diffusion with Differentiable Noisy Historycited by 0
  30. 2026LongTake: Learning to Sustain Dynamics in Long-Horizon Video Generationcited by 0
  31. 2026Memorizon: Training World Models Beyond Their Context Windowcited by 0
  32. 2026SGF+: Decoupling Gradient Flows for Autoregressive Video Generationcited by 0
  33. 2026Conditional Residual Prediction: Improving Autoregressive Video Diffusion without a Bidirectional Teachercited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Self Gradient Forcing: Native Long Video Extrapolation
  2. 2026In-Context Forcing: Uncovering Context Effects in Autoregressive Video Diffusion
  3. 2026Causal-rCM: A Unified Teacher-Forcing and Self-Forcing Open Recipe for Autoregressive Diffusion Distillation in Streaming Video Generation and Interactive World Models
  4. 2026Rolling Sink: Bridging Limited-Horizon Training and Open-Ended Testing in Autoregressive Video Diffusion
  5. 2025BAgger: Backwards Aggregation for Mitigating Drift in Autoregressive Video Diffusion Models
  6. 2025Playing with Transformer at 30+ FPS via Next-Frame Diffusion
  7. 2026OPSD-V: On-Policy Self-Distillation for Post-Training Few-Step Autoregressive Video Generators
  8. 2026Video-Mirai: Autoregressive Video Diffusion Models Need Foresight