Decide Then Retrieve: A Training-Free Framework with Uncertainty-Guided Triggering and Dual-Path Retrieval
This work addresses noise and performance limitations in RAG systems for open-domain QA, offering an incremental improvement with adaptive retrieval mechanisms.
The paper tackles the problem of indiscriminate retrieval and single-path evidence construction in retrieval-augmented generation (RAG) for large language models, proposing a training-free framework called Decide Then Retrieve (DTR) that uses uncertainty-guided triggering and dual-path retrieval to improve performance and reduce unnecessary retrievals, achieving consistent gains in EM and F1 across multiple benchmarks.
Retrieval-augmented generation (RAG) enhances large language models (LLMs) by incorporating external knowledge, but existing approaches indiscriminately trigger retrieval and rely on single-path evidence construction, often introducing noise and limiting performance gains. In this work, we propose Decide Then Retrieve (DTR), a training-free framework that adaptively determines when retrieval is necessary and how external information should be selected. DTR leverages generation uncertainty to guide retrieval triggering and introduces a dual-path retrieval mechanism with adaptive information selection to better handle sparse and ambiguous queries. Extensive experiments across five open-domain QA benchmarks, multiple model scales, and different retrievers demonstrate that DTR consistently improves EM and F1 over standard RAG and strong retrieval-enhanced baselines, while reducing unnecessary retrievals. The code and data used in this paper are available at https://github.com/ChenWangHKU/DTR.