Search papers, labs, and topics across Lattice.
Tencent
3
0
6
5
Relevance-aware interaction can drastically accelerate search convergence, revealing critical information sooner in complex queries.
Zero-shot learning can now handle noisy, ambiguous labels thanks to a CLIP-driven framework that progressively refines labels during training.
Achieve state-of-the-art long-context reranking with a lightweight (4B parameter) model by cleverly focusing on query-relevant attention heads, outperforming much larger models.
