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

TL;DR

LACONIC introduces a scalable, efficient learned sparse retrieval method based on Llama-3, achieving competitive performance with reduced memory and computational requirements through a novel two-phase training curriculum.

Contribution

The paper presents LACONIC, a new sparse retrieval approach using Llama-3 models and a two-phase training process, bridging the performance gap with dense models while reducing resource usage.

Findings

Achieves 60.2 nDCG on MTEB Retrieval benchmark.

Uses 71% less index memory than dense models.

Operates effectively on commodity CPU hardware.

Abstract

While dense retrieval models have become the standard for state-of-the-art information retrieval, their deployment is often constrained by high memory requirements and reliance on GPU accelerators for vector similarity search. Learned sparse retrieval offers a compelling alternative by enabling efficient search via inverted indices, yet it has historically received less attention than dense approaches. In this report, we introduce LACONIC, a family of learned sparse retrievers based on the Llama-3 architecture (1B, 3B, and 8B). We propose a streamlined two-phase training curriculum consisting of (1) weakly supervised pre-finetuning to adapt causal LLMs for bidirectional contextualization and (2) high-signal finetuning using curated hard negatives. Our results demonstrate that LACONIC effectively bridges the performance gap with dense models: the 8B variant achieves a state-of-the-art 60.2 nDCG on the MTEB Retrieval benchmark, ranking 15th on the leaderboard as of January 1, 2026, while utilizing 71\% less index memory than an equivalent dense model. By delivering high retrieval effectiveness on commodity CPU hardware with a fraction of the compute budget required by competing models, LACONIC provides a scalable and efficient solution for real-world search applications.