Decoupled Transformer for Scalable Inference in Open-domain Question Answering
This addresses the problem of scalable inference for online QA systems like voice assistants, offering a practical improvement but is incremental in nature.
The paper tackled the high computational cost and latency of transformer models in open-domain question answering by proposing a decoupled transformer that splits the model into input and cross components, enabling offline caching and online efficiency. The result was a 30-40% reduction in computational cost and latency with only a 1.2-point drop in F1-score on the SQUAD 2.0 dataset.
Large transformer models, such as BERT, achieve state-of-the-art results in machine reading comprehension (MRC) for open-domain question answering (QA). However, transformers have a high computational cost for inference which makes them hard to apply to online QA systems for applications like voice assistants. To reduce computational cost and latency, we propose decoupling the transformer MRC model into input-component and cross-component. The decoupling allows for part of the representation computation to be performed offline and cached for online use. To retain the decoupled transformer accuracy, we devised a knowledge distillation objective from a standard transformer model. Moreover, we introduce learned representation compression layers which help reduce by four times the storage requirement for the cache. In experiments on the SQUAD 2.0 dataset, a decoupled transformer reduces the computational cost and latency of open-domain MRC by 30-40% with only 1.2 points worse F1-score compared to a standard transformer.