Search papers, labs, and topics across Lattice.
6
0
9
13
Adversarial documents generated by VerTox can out-rank legitimate content, exposing critical vulnerabilities in neural ranking models that power modern AI systems.
Adaptive self-distillation can boost model performance by over 23 points without extra rollouts, reshaping how we approach teacher-student dynamics in training.
Bridging the semantic gap between images and text can dramatically enhance multimodal RAG performance, as shown by our novel Context-Enhanced MMKG framework.
A single checkpoint can now adapt to any model size, streamlining the deployment of elastic retrieval systems and achieving faster performance without sacrificing quality.
Achieving high-quality reranking with a 30B MoE model is now feasible on an academic budget, outperforming traditional dense models in efficiency.
Last-utterance clarifications can worsen parsing accuracy, with parser-agnostic rewrites introducing more errors than fixes in real-world applications.
