Conformer-Kernel with Query Term Independence for Document Retrieval

The Transformer-Kernel (TK) model has demonstrated strong reranking performance on the TREC Deep Learning benchmark—and can be considered to be an efficient (but slightly less effective) alternative to BERT-based ranking models. In this work, we extend the TK architecture to the full retrieval setting by incorporating the query term independence assumption. Furthermore, to reduce the memory complexity of the Transformer layers with respect to the input sequence length, we propose a new Conformer layer. We show that the Conformer’s GPU memory requirement scales linearly with input sequence length, making it a more viable option when ranking long documents. Finally, we demonstrate that incorporating explicit term matching signal into the model can be particularly useful in the full retrieval setting. We present preliminary results from our work in this paper.

Publication Downloads

Conformer-Kernel Model with Query Term Independence (TREC Deep Learning Quick Start)

March 16, 2021

This is a quick start guide for the document ranking task in the TREC Deep Learning (TREC-DL) benchmark. If you are new to TREC-DL, then this repository may make it more convenient for you to download all the required datasets and then train and evaluate a relatively efficient deep neural baseline on this benchmark, under both the rerank and the fullrank settings. The base model implements the Conformer-Kernel architecture with QTI, as described in this paper.