gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion
From diffusion forcing to speculative decoding: a living reading list
A curated bibliography sorting AR diffusion video world models into algorithms, applications, and the infrastructure required for real-time interaction.
Not currently ranked — collecting fresh signals.
star history
What it does This repository is a curated awesome list tracking research on autoregressive diffusion for video world modeling. It catalogs papers and projects across three axes—algorithmic foundations, real-world applications, and infrastructure-level acceleration—and bundles the lot into a single downloadable BibTeX file.
The interesting bit Most literature surveys stop at the model architecture; this one treats inference infrastructure as a first-class citizen, giving sparse attention, caching, quantized attention, and speculative decoding their own sections. The maintainers also eat their own cooking, building and promoting a companion full-stack open-source framework called minWM for turning text-to-video models into action-conditioned world models.
Key highlights
- Taxonomy spans native pretraining, distillation for real-time generation, long-video methods, embodied AI, and avatar control.
- Ships a consolidated
video-world-models.bibfor one-click import into LaTeX or Zotero. - Explicitly covers “AR Diffusion Distillation” (e.g., CausVid, Self Forcing, Causal Forcing) as a distinct category for streaming generation.
- Updated weekly, according to the README.
- Curated by active researchers who also publish in the space (see minWM).
Caveats
- The authors acknowledge the list is not yet exhaustive and actively welcome pull requests for omissions.
- Beyond categorization and BibTeX links, there is little original synthesis or commentary on the papers themselves.
Verdict Worth bookmarking if you are actively researching or engineering video world models, especially where generative diffusion meets real-time interaction. Skip it if you need a turnkey SDK or consumer application—this is a reading list, not a product.
Frequently asked
- What is gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion?
- A curated bibliography sorting AR diffusion video world models into algorithms, applications, and the infrastructure required for real-time interaction.
- Is Awesome-Video-World-Models-with-AR-Diffusion open source?
- Yes — gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion is an open-source project tracked on heatdrop.
- What language is Awesome-Video-World-Models-with-AR-Diffusion written in?
- gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion is primarily written in TeX.
- How popular is Awesome-Video-World-Models-with-AR-Diffusion?
- gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion has 666 stars on GitHub.
- Where can I find Awesome-Video-World-Models-with-AR-Diffusion?
- gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion is on GitHub at https://github.com/gracezhao1997/Awesome-Video-World-Models-with-AR-Diffusion.