Search papers, labs, and topics across Lattice.
3
0
6
PriFT achieves state-of-the-art performance in supervised fine-tuning by leveraging a stable token reweighting signal from a frozen pretrained model, drastically improving generalization.
Training VLA policies without human demos is now feasible, with LEGS achieving better performance than traditional methods at a fraction of the cost.
Tencent's SIREN model boosts ad revenue by up to 3.87% in Weixin by unifying multi-modal and collaborative data into a single transformer, outperforming traditional late-fusion approaches.
