LACONIC: Dense-Level Effectiveness for Scalable Sparse Retrieval via a Two-Phase Training Curriculum

By Zhi-Chao Xu, Shengyao Zhuang, Crystina Zhang, Xue-Guang Ma, Yi-Jun Tian, Maitrey Mehta, Jimmy Lin, Vivek Srikumar. 2026

Subfields: Information Retrieval & Search, Representation & Self-Supervised Learning

Citations in the corpus

• Mixture-of-Experts Language Models Can Be Strong and Efficient Retrievers (2026, unknown): “Across several model families, larger backbones generally yield stronger retrievers (Ni et al., 2022; Zhang et al., 2025; Xu et al., 2026b).”