Large Vision-Language Models (LVLMs) face significant computational inefficiencies caused by the large number of visual tokens. Existing visual token pruning methods mainly focus on either retaining individually important tokens or selecting mutually diverse ones. In this work, we revisit visual token pruning from a co...
A lightweight quality gate identifies coverage gaps in initial keyword results, adaptively routing only the ∼ 10% of users who benefit from semantic prospecting to the LLM path, and captures the bulk of semantic prospecting gains in Relevance Score while maintaining competitive retargeting performance at a 90% reductio...
Cong-Fei Zhang, J. Ma, Xiao-Dong Liu et al.· Proceedings of the 20th ACM...· 0 citations
Deployed as a new retrieval source in the DPA production stack, SetMIR lifts overall CVR by 3.1%, while lifting CTR by 44% and CVR by 51% over the item-to-item retrieval source with the same item embeddings, ANN index, and retrieval quota.
Xiao-Dong Liu, Cong-Fei Zhang, Hsiang-Wei Chao et al.· 0 citations
CRISP is proposed, a pre-LLM yet text-driven visual token pruning framework that preserves both instruction-relevant evidence and essential scene context that serves as a practical solution for efficient LVLM inference, especially in resource-constrained scenarios.
Xu Li, Yi Zheng, Mengyang Zhao et al.· arXiv.org· 1 citation
EGR is proposed, an Embedding-Native Generative Retrieval framework that uses a single shared LLM to learn item representations from item metadata and user representations from interaction histories in one embedding space, simplifying system design while improving retrieval quality and ad performance.
