Search papers, labs, and topics across Lattice.
M 67.80±0.4567.80{\scriptstyle\pm 0.45} 83.25±0.3083.25{\scriptstyle\pm 0.30} 72.99±0.5072.99{\scriptstyle\pm 0.50} 86.74±0.3386.74{\scriptstyle\pm 0.33} FewTURE [44] NeurIPS’22 ViT-S/16 22.
1
4
3
14
LLMs can generate semantically consistent images and textual descriptions that, when geometrically aligned, dramatically improve few-shot learning performance.