Efficient Title Reranker for Fast and Improved Knowledge-Intense NLP
This work addresses efficiency bottlenecks in knowledge-intensive NLP for researchers and practitioners, though it is incremental as it builds on existing reranking methods.
The paper tackled the high computational cost of existing rerankers in RAG approaches by introducing an efficient title reranker with a broadcasting query encoder and a novel loss function, achieving a 20x-40x speedup and state-of-the-art results on four KILT benchmark datasets.
In recent RAG approaches, rerankers play a pivotal role in refining retrieval accuracy with the ability of revealing logical relations for each pair of query and text. However, existing rerankers are required to repeatedly encode the query and a large number of long retrieved text. This results in high computational costs and limits the number of retrieved text, hindering accuracy. As a remedy of the problem, we introduce the Efficient Title Reranker via Broadcasting Query Encoder, a novel technique for title reranking that achieves a 20x-40x speedup over the vanilla passage reranker. Furthermore, we introduce Sigmoid Trick, a novel loss function customized for title reranking. Combining both techniques, we empirically validated their effectiveness, achieving state-of-the-art results on all four datasets we experimented with from the KILT knowledge benchmark.