Fetching the paper…

iFlame: Interleaving Full and Linear Attention for Efficient Mesh Generation · Around