Can Large Language Models Recall Reference Location Like Humans?
Wang, Ye, Xu, Xinrun, Xie, Rui, Hu, Wenxin, Ye, Wei
–arXiv.org Artificial Intelligence
When completing knowledge-intensive tasks, humans sometimes need not just an answer but also a corresponding reference passage for auxiliary reading. Previous methods required obtaining pre-segmented article chunks through additional retrieval models. This paper explores leveraging the parameterized knowledge stored during the pre-training phase of large language models (LLMs) to independently recall reference passage from any starting position. We propose a two-stage framework that simulates the scenario of humans recalling easily forgotten references. Initially, the LLM is prompted to recall document title identifiers to obtain a coarse-grained document set. Then, based on the acquired coarse-grained document set, it recalls fine-grained passage. In the two-stage recall process, we use constrained decoding to ensure that content outside of the stored documents is not generated. To increase speed, we only recall a short prefix in the second stage, then locate its position to retrieve a complete passage. Experiments on KILT knowledge-sensitive tasks have verified that LLMs can independently recall reference passage location in various task forms, and the obtained reference significantly assist downstream tasks.
arXiv.org Artificial Intelligence
Feb-26-2024
- Country:
- Oceania > Australia (0.14)
- North America
- United States > Louisiana
- Orleans Parish > New Orleans (0.04)
- Canada > Ontario
- Toronto (0.04)
- United States > Louisiana
- Europe
- Greece (0.05)
- United Kingdom (0.04)
- Norway (0.04)
- Italy (0.04)
- Belgium > Brussels-Capital Region
- Brussels (0.04)
- Asia
- South Korea (0.04)
- China (0.04)
- Middle East > UAE
- Abu Dhabi Emirate > Abu Dhabi (0.04)
- Africa
- North Africa (0.04)
- South Africa > Western Cape
- Cape Town (0.04)
- Eswatini > Manzini
- Manzini (0.04)
- Genre:
- Research Report (0.82)
- Industry:
- Media (1.00)
- Leisure & Entertainment (1.00)
- Law (0.93)
- Government > Military (0.93)
- Technology: