Transformers today still struggle to generate one-minute videos because self-attention layers are inefficient for long context. Alternatives such as Mamba layers struggle with complex multi-scene stories because their hidden states are less expressive. We experiment with Test-Time Training (TTT) layers, whose hidden states themselves can be neural networks, therefore more expressive. Adding TTT layers into a pre-trained Transformer enables it to generate one-minute videos from text storyboards. For proof of concept, we curate a dataset based on Tom and Jerry cartoons. Compared to baselines such as Mamba~2, Gated DeltaNet, and sliding-window attention layers, TTT layers generate much more coherent videos that tell complex stories, leading by 34 Elo points in a human evaluation of 100 videos per method. Although promising, results still contain artifacts, likely due to the limited capability of the pre-trained 5B model. The efficiency of our implementation can also be improved. We have only experimented with one-minute videos due to resource constraints, but the approach can be extended to longer videos and more complex stories. Sample videos, code and annotations are available at:this https URL
View on arXiv@article{dalal2025_2504.05298,
title={ One-Minute Video Generation with Test-Time Training },
author={ Karan Dalal and Daniel Koceja and Gashon Hussein and Jiarui Xu and Yue Zhao and Youjin Song and Shihao Han and Ka Chun Cheung and Jan Kautz and Carlos Guestrin and Tatsunori Hashimoto and Sanmi Koyejo and Yejin Choi and Yu Sun and Xiaolong Wang },
journal={arXiv preprint arXiv:2504.05298},
year={ 2025 }
}