Autoregressive Video Generation without Vector Quantization

Haoge Deng, Ting Pan, Haiwen Diao, Zhengxiong Luo, Yufeng Cui +4 more
2/12/2026

No official code found. Best implementation: baaivision/NOVA (646 stars, Apache-2.0, last commit 2025-10-29).

Semantic Scholar

Code Implementations(1)

baaivision/NOVACommunity76%
Apache-2.0

[ICLR 2025] Autoregressive Video Generation without Vector Quantization

64622PythonDec 16, 202411 months agoApache-2.0
autoregressive-modelsdiffusion-modelsimage-generationvideo-generation

Cite this paper

@article{deng2026autoregressive,
  title  = {Autoregressive Video Generation without Vector Quantization},
  author = {Haoge Deng and Ting Pan and Haiwen Diao and Zhengxiong Luo and Yufeng Cui and Huchuan Lu and Shiguang Shan and Yonggang Qi and Xinlong Wang},
  year   = {2026},
  url    = {https://api.semanticscholar.org/CorpusID:980f09eb9a16221ce3ebe1b1471f0e6c671d7734},
  journal = {ICLR 2025 2025}
}

Discussion