Search papers, labs, and topics across Lattice.
5
2
7
9
RGD reshapes the decoding process in generative recommendation, ensuring high-value candidates are prioritized without retraining the model.
A unified model that seamlessly integrates generative recall and multi-objective ranking achieves significant performance improvements in real-world recommendation systems.
Learned attention allocation patterns reveal that SWA is best positioned in lower layers, challenging conventional wisdom on attention distribution in LLMs.
SuperFashion achieves up to 9.35% improvement in fashion image retrieval by leveraging superpixel tokens for enhanced attribute localization.
Forget one-size-fits-all recommendations: this model uses normalizing flows to capture the *multimodal* nature of individual user preferences, leading to better cold-start performance in cross-domain recommendation.
