Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2504.10816 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908634783416320 |
|---|---|
| author | Xu, Zhichao Feng, Aosong Tian, Yijun Ding, Haibo Cheong, Lin Lee |
| author_facet | Xu, Zhichao Feng, Aosong Tian, Yijun Ding, Haibo Cheong, Lin Lee |
| contents | In recent years, dense retrieval has been the focus of information retrieval (IR) research. While effective, dense retrieval produces uninterpretable dense vectors, and suffers from the drawback of large index size. Learned sparse retrieval (LSR) has emerged as promising alternative, achieving competitive retrieval performance while also being able to leverage the classical inverted index data structure for efficient retrieval. However, limited works have explored scaling LSR beyond BERT scale. In this work, we identify two challenges in training large language models (LLM) for LSR: (1) training instability during the early stage of contrastive training; (2) suboptimal performance due to pre-trained LLM's unidirectional attention. To address these challenges, we propose two corresponding techniques: (1) a lightweight adaptation training phase to eliminate training instability; (2) two model variants to enable bidirectional information. With these techniques, we are able to train LSR models with 8B scale LLM, and achieve competitive retrieval performance with reduced index size. Furthermore, we are among the first to analyze the performance-efficiency tradeoff of LLM-based LSR model through the lens of model quantization. Our findings provide insights into adapting LLMs for efficient retrieval modeling. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_10816 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CSPLADE: Learned Sparse Retrieval with Causal Language Models Xu, Zhichao Feng, Aosong Tian, Yijun Ding, Haibo Cheong, Lin Lee Information Retrieval Computation and Language In recent years, dense retrieval has been the focus of information retrieval (IR) research. While effective, dense retrieval produces uninterpretable dense vectors, and suffers from the drawback of large index size. Learned sparse retrieval (LSR) has emerged as promising alternative, achieving competitive retrieval performance while also being able to leverage the classical inverted index data structure for efficient retrieval. However, limited works have explored scaling LSR beyond BERT scale. In this work, we identify two challenges in training large language models (LLM) for LSR: (1) training instability during the early stage of contrastive training; (2) suboptimal performance due to pre-trained LLM's unidirectional attention. To address these challenges, we propose two corresponding techniques: (1) a lightweight adaptation training phase to eliminate training instability; (2) two model variants to enable bidirectional information. With these techniques, we are able to train LSR models with 8B scale LLM, and achieve competitive retrieval performance with reduced index size. Furthermore, we are among the first to analyze the performance-efficiency tradeoff of LLM-based LSR model through the lens of model quantization. Our findings provide insights into adapting LLMs for efficient retrieval modeling. |
| title | CSPLADE: Learned Sparse Retrieval with Causal Language Models |
| topic | Information Retrieval Computation and Language |
| url | https://arxiv.org/abs/2504.10816 |