
    \i                          d Z ddlmZ dgZy)zrVideo latent prediction model: DINO first-frame + learnable patches, self-attention, conv decode to 32x32 targets.   )VideoLatentModelr   N)__doc__modelr   __all__     ./data/cameron/vidgen/our_vid_model/__init__.py<module>r
      s    x #
r   