j1s4nn/f4lite-videos

F4-Lite Adaptive Granularity Rollout Videos

T4 online-policy rollout videos for adaptive temporal granularity

F4-Lite Adaptive Granularity Rollout Videos

The original F4-Lite online-policy study for adaptive temporal granularity in causal video generation. Rollout videos generated by Causal-rCM (Wan2.1-T2V-1.3B) on a Tesla T4, 8 fresh prompts, 57 frames, comparing fixed coarse, fixed fine, random-50, alternating-50, adaptive-warp policies, and a reference.

Related resources

• Code (JevGAG programme): https://github.com/j1s4nn/JevGAG
• Code (Jev-Routed Temporal Granularity): https://github.com/j1s4nn/jev-adaptive-granularity-WM
• Project page (JevGAG): https://j1s4nn.github.io/JevGAG/
• Project page (Paper 1): https://j1s4nn.github.io/jev-adaptive-granularity-WM/
• Portfolio: https://j1s4nn.github.io
• Source (Kaggle): https://www.kaggle.com/datasets/ajjisan/f4lite-videos
• Preprint (JevGAG): https://doi.org/10.5281/zenodo.23693517
• Preprint (Jev-Routed Temporal Granularity): https://doi.org/10.5281/zenodo.23395414

Citation

@misc{hossen_jevgag_ddag,
  author       = {Hossen, MD Jisan},
  title        = {Regime-Aware Temporal Granularity for Causal Video World Models: When Does Adaptivity Pay?},
  year         = {2026},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.23693517},
  url          = {https://doi.org/10.5281/zenodo.23693517}
}

License

CC-BY-4.0. Maintained by Vision Intelligence (MD Jisan Hossen (NUIST)).