Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2504.11420 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866915244129910784 |
|---|---|
| author | Long, Quanyu Chen, Jianda Liu, Zhengyuan Chen, Nancy F. Wang, Wenya Pan, Sinno Jialin |
| author_facet | Long, Quanyu Chen, Jianda Liu, Zhengyuan Chen, Nancy F. Wang, Wenya Pan, Sinno Jialin |
| contents | Large Language Models (LLMs) have demonstrated remarkable capabilities across numerous tasks, yet they often rely on external context to handle complex tasks. While retrieval-augmented frameworks traditionally focus on selecting top-ranked documents in a single pass, many real-world scenarios demand compositional retrieval, where multiple sources must be combined in a coordinated manner. In this work, we propose a tri-encoder sequential retriever that models this process as a Markov Decision Process (MDP), decomposing the probability of retrieving a set of elements into a sequence of conditional probabilities and allowing each retrieval step to be conditioned on previously selected examples. We train the retriever in two stages: first, we efficiently construct supervised sequential data for initial policy training; we then refine the policy to align with the LLM's preferences using a reward grounded in the structural correspondence of generated programs. Experimental results show that our method consistently and significantly outperforms baselines, underscoring the importance of explicitly modeling inter-example dependencies. These findings highlight the potential of compositional retrieval for tasks requiring multiple pieces of evidence or examples. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_11420 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Reinforcing Compositional Retrieval: Retrieving Step-by-Step for Composing Informative Contexts Long, Quanyu Chen, Jianda Liu, Zhengyuan Chen, Nancy F. Wang, Wenya Pan, Sinno Jialin Computation and Language Large Language Models (LLMs) have demonstrated remarkable capabilities across numerous tasks, yet they often rely on external context to handle complex tasks. While retrieval-augmented frameworks traditionally focus on selecting top-ranked documents in a single pass, many real-world scenarios demand compositional retrieval, where multiple sources must be combined in a coordinated manner. In this work, we propose a tri-encoder sequential retriever that models this process as a Markov Decision Process (MDP), decomposing the probability of retrieving a set of elements into a sequence of conditional probabilities and allowing each retrieval step to be conditioned on previously selected examples. We train the retriever in two stages: first, we efficiently construct supervised sequential data for initial policy training; we then refine the policy to align with the LLM's preferences using a reward grounded in the structural correspondence of generated programs. Experimental results show that our method consistently and significantly outperforms baselines, underscoring the importance of explicitly modeling inter-example dependencies. These findings highlight the potential of compositional retrieval for tasks requiring multiple pieces of evidence or examples. |
| title | Reinforcing Compositional Retrieval: Retrieving Step-by-Step for Composing Informative Contexts |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2504.11420 |