mlsanityData to 7 Oct 2026

Taming Teacher Forcing for Masked Autoregressive Video Generation

arXiv 2501.12389 ↗First posted 21 Jan 2025cs.CVcited by 29

Not on the map. Why?

Abstract (arXiv)

We introduce MAGI, a hybrid video generation framework that combines masked modeling for intra-frame generation with causal modeling for next-frame generation. Our key innovation, Complete Teacher Forcing (CTF), conditions masked frames on complete observation frames rather than masked ones (namely Masked Teacher Forcing, MTF), enabling a smooth transition from token-level (patch-level) to frame-level autoregressive generation. CTF significantly outperforms MTF, achieving a +23% improvement in FVD scores on first-frame conditioned video prediction. To address issues like exposure bias, we employ targeted training strategies, setting a new benchmark in autoregressive video generation. Experiments show that MAGI can generate long, coherent video sequences exceeding 100 frames, even when trained on as few as 16 frames, highlighting its potential for scalable, high-quality video generation.

Key facts. Cited by 29 papers in the map (data through 7 Oct 2026). Builds on 48 of its 58 references that are in the map. First posted on arXiv 21 Jan 2025.

Builds on

48 of 58 references.

References · 48

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  3. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  4. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,339
  5. 2022Classifier-Free Diffusion Guidancecited by 5,717
  6. 2017Neural Discrete Representation Learningcited by 4,650
  7. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,290
  8. 2021Zero-Shot Text-to-Image Generationcited by 4,276
  9. 2022Video Diffusion Modelscited by 2,280
  10. 2023AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuningcited by 1,418
  11. 2018Towards Accurate Generative Models of Video: A New Metric & Challengescited by 1,204
  12. 2022MaskGIT: Masked Generative Image Transformercited by 1,082
36 more
  1. 2024WorldSimBench: Towards Video Generation Models as World Simulatorscited by 867
  2. 2023PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesiscited by 854
  3. 2024Emu3: Next-Token Prediction is All You Needcited by 761
  4. 2024Genie: Generative Interactive Environmentscited by 706
  5. 2021VideoGPT: Video Generation using VQ-VAE and Transformerscited by 651
  6. 2024Autoregressive Image Generation without Vector Quantizationcited by 644
  7. 2023Muse: Text-To-Image Generation via Masked Generative Transformerscited by 640
  8. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 634
  9. 2023Language Model Beats Diffusion -- Tokenizer is Key to Visual Generationcited by 633
  10. 2024LTX-Video: Realtime Video Latent Diffusioncited by 549
  11. 2024Latte: Latent Diffusion Transformer for Video Generationcited by 536
  12. 2017MoCoGAN: Decomposing Motion and Content for Video Generationcited by 523
  13. 2023VideoPoet: A Large Language Model for Zero-Shot Video Generationcited by 479
  14. 2022Phenaki: Variable Length Video Generation From Open Domain Textual Descriptioncited by 475
  15. 2023Model Tells You What to Discard: Adaptive KV Cache Compression for LLMscited by 466
  16. 2022MAGVIT: Masked Generative Video Transformercited by 341
  17. 2018A Short Note about Kinetics-600cited by 331
  18. 2021StyleGAN-V: A Continuous Video Generator with the Price, Image Quality and Perks of StyleGAN2cited by 300
  19. 2024Diffusion Models Are Real-Time Game Enginescited by 282
  20. 2021N"UWA: Visual Synthesis Pre-training for Neural visUal World creAtioncited by 253
  21. 2016Professor Forcing: A New Algorithm for Training Recurrent Networkscited by 214
  22. 2024How Far is Video Generation from World Model: A Physical Law Perspectivecited by 204
  23. 2022Scalable Adaptive Computation for Iterative Generationcited by 190
  24. 2017DART: Noise Injection for Robust Imitation Learningcited by 187
  25. 2022Generating Videos with Dynamics-aware Implicit Generative Adversarial Networkscited by 186
  26. 2023Video Probabilistic Diffusion Models in Projected Latent Spacecited by 171
  27. 2021A Good Image Generator Is What You Need for High-Resolution Video Synthesiscited by 162
  28. 2024Fluid: Scaling Autoregressive Text-to-image Generative Models with Continuous Tokenscited by 147
  29. 2019Bridging the Gap between Training and Inference for Neural Machine Translationcited by 123
  30. 2019Generalization in Generation: A closer look at Exposure Biascited by 123
  31. 2024OmniTokenizer: A Joint Image-Video Tokenizer for Visual Generationcited by 117
  32. 2024Keep the Cost Down: A Review on Methods to Optimize LLM' s KV-Cache Consumptioncited by 96
  33. 2024DiM: Diffusion Mamba for Efficient High-Resolution Image Synthesiscited by 72
  34. 2021Predicting Video with VQVAEcited by 61
  35. 2023MoStGAN-V: Video Generation with Temporal Motion Stylescited by 37
  36. 2024Scaling Diffusion Mamba with Bidirectional SSMs for Efficient Image and Video Generationcited by 21

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 18; 2026: 11.

29 papers.

Sort:

By area:

  1. 2025Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusioncited by 569
  2. 2025LongLive: Real-time Interactive Long Video Generationcited by 237
  3. 2025Step-Video-T2V Technical Report: The Practice, Challenges, and Future of Video Foundation Modelcited by 163
  4. 2025Long-Context Autoregressive Video Modeling with Next-Frame Predictioncited by 119
  5. 2025Mogao: An Omni Foundation Model for Interleaved Multi-Modal Generationcited by 108
  6. 2025RELIC: Interactive Video World Model with Long-Horizon Memorycited by 89
  7. 2026LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generationcited by 25
  8. 2025Lumos-1: On Autoregressive Video Generation with Discrete Diffusion from a Unified Model Perspectivecited by 22
  9. 2026Relax Forcing: Relaxed KV-Memory for Consistent Long Video Generationcited by 20
  10. 2025Generative Pre-trained Autoregressive Diffusion Transformercited by 16
  11. 2025VideoMAR: Autoregressive Video Generatio with Continuous Tokenscited by 15
  12. 2025WorldWeaver: Generating Long-Horizon Video Worlds via Rich Perceptioncited by 14
17 more
  1. 2025MaskFlow: Discrete Flows For Flexible and Efficient Long Video Generationcited by 12
  2. 2025Video-GPT via Next Clip Diffusioncited by 8
  3. 2025REAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularizationcited by 6
  4. 2025SemanticGen: Video Generation in Semantic Spacecited by 6
  5. 2026Context-Matched Distillation: Teacher Causality for Autoregressive Video Distillationcited by 5
  6. 2025SAMPO:Scale-wise Autoregression with Motion PrOmpt for generative world modelscited by 4
  7. 2025Generative Augmented Reality: Paradigms, Technologies, and Future Applicationscited by 1
  8. 2026AnchorEdit: Maintaining Temporal Consistency in Multi-turn Image Editing via Causal Memorycited by 1
  9. 2025CanvasMAR: Improving Masked Autoregressive Video Prediction With Canvascited by 0
  10. 2025Real-Time Long Horizon Air Quality Forecasting via Group-Relative Policy Optimizationcited by 0
  11. 2026Entropy-Guided k-Guard Sampling for Long-Horizon Autoregressive Video Generationcited by 0
  12. 2026GDPO-Listener: Expressive Interactive Head Generation via Auto-Regressive Flow Matching and Group reward-Decoupled Policy Optimizationcited by 0
  13. 2026Real-Time Generation of Streamable Talking Portrait Video with Reference-Guided Deep Compression VAEscited by 0
  14. 2026Vorch-IR: Long-Form Unified Multimodal Identity Replacement Video Generationcited by 0
  15. 2026SNF-Bench: Separating Static Drift from Natural Flow in Long-Horizon Fixed-Camera Video Generationcited by 0
  16. 2026Temporal Gradient Inversion for Private Trajectory Reconstruction in Embodied Reinforcement Learningcited by 0
  17. 2026Unveiling the Value of Motion for Cinematic Camera Trajectoriescited by 0