Train Short, Inference Long: Training-free Horizon Extension for Autoregressive Video Generation

arXiv:2602.14027v2 Announce Type: replace Abstract: Autoregressive video diffusion models have emerged as a scalable paradigm for long video generation. However, they often suffer from severe extrapolation failure, where rapid error accumulation leads to significant temporal d...

🔗 Read more: https://arxiv.org/abs/2602.14027

#News #Chemistry #Energy #WorldNews #Policy #Biology #Academic