InRanker: Distilled Rankers for Zero-Shot Information Retrieval
Resumo
Despite multi-billion parameter neural rankers being common components of state-of-the-art information retrieval pipelines, they are rarely used in production due to the enormous amount of compute required for inference. In this work, we propose a method for distilling large rankers into their smaller versions focusing on out-of-domain effectiveness. We introduce InRanker, a version of monoT5 [25] distilled from monoT5-3B with increased effectiveness on out-of-domain scenarios. Our key insight is to use language models and rerankers to generate as much as possible synthetic “in-domain” training data, i.e., data that closely resembles the data that will be seen at retrieval time. The pipeline consists of two distillation phases that do not require additional user queries or manual annotations: (1) training on existing supervised soft teacher labels, and (2) training on teacher soft labels for synthetic queries generated using a large language model. Consequently, models like monoT5-60M and monoT5-220M improved their effectiveness by using the teacher’s knowledge, despite being 50x and 13x smaller, respectively. Furthermore, we show that it is possible to transfer knowledge from English models to Portuguese fine-tuned models. Models and code are available at https://github.com/unicamp-dl/inranker.
