Curriculum Learning For Dense Retrieval Distillation

Abstract

Recent work has shown that more effective dense retrieval models can be obtained by distilling ranking knowledge from an existing base re-ranking model. In this paper, we propose a generic curriculum learning based optimization framework called CL-DRD that controls the difficulty level of training data produced by the re-ranking (teacher) model. CL-DRD iteratively optimizes the dense retrieval (student) model by increasing the difficulty of the knowledge distillation data made available to it. In more detail, we initially provide the student model coarse-grained preference pairs between documents in the teacher's ranking and progressively move towards finer-grained pairwise document ordering requirements. In our experiments, we apply a simple implementation of the CL-DRD framework to enhance two state-of-the-art dense retrieval models. Experiments on three public passage retrieval datasets demonstrate the effectiveness of our proposed framework.

Stats

• citations33
• S2 citations—
• github stars0
• HF likes0
• heat score11.49
• arxiv keyzeng2022curriculum

Related papers

• PROD: Progressive Distillation For Dense Retrieval (2022)9.23
• Pairdistill: Pairwise Relevance Distillation For Dense Retrieval (2024)7.24
• Learning To Retrieve: How To Train A Dense Retrieval Model Effectively And Efficiently (2020)0.00
• Teaching Dense Retrieval Models To Specialize With Listwise Distillation And LLM Data Augmentation (2025)0.00
• Data-efficient Ranking Distillation For Image Retrieval (2020)0.00
• Towards Dynamic Dense Retrieval With Routing Strategy (2026)0.00
• Knowledge Distillation In Document Retrieval (2019)0.00
• Translate-distill: Learning Cross-language Dense Retrieval By Translation And Distillation (2024)8.60