Towards Dynamic Dense Retrieval with Routing Strategy
This work addresses the high cost and data dependency of adapting dense retrieval models for new domains, which is a problem for researchers and practitioners in information retrieval.
The paper proposes Dynamic Dense Retrieval (DDR) to overcome the limitations of fine-tuning dense retrieval models for new tasks, especially with limited data or frequent updates. DDR uses prefix tuning modules for domain adaptation, which are then combined using a dynamic routing strategy. This approach achieves better performance than traditional dense retrieval while using only 2% of the training parameters on six zero-shot downstream tasks.
The \textit{de facto} paradigm for applying dense retrieval (DR) to new tasks involves fine-tuning a pre-trained model for a specific task. However, this paradigm has two significant limitations: (1) It is difficult adapt the DR to a new domain if the training dataset is limited. (2) Old DR models are simply replaced by newer models that are trained from scratch when the former are no longer up to date. Especially for scenarios where the model needs to be updated frequently, this paradigm is prohibitively expensive. To address these challenges, we propose a novel dense retrieval approach, termed \textit{dynamic dense retrieval} (DDR). DDR uses \textit{prefix tuning} as a \textit{module} specialized for a specific domain. These modules can then be compositional combined with a dynamic routing strategy, enabling highly flexible domain adaptation in the retrieval part. Extensive evaluation on six zero-shot downstream tasks demonstrates that this approach can surpass DR while utilizing only 2\% of the training parameters, paving the way to achieve more flexible dense retrieval in IR. We see it as a promising future direction for applying dense retrieval to various tasks.