An Efficient Memory-Augmented Transformer for Knowledge-Intensive NLP Tasks
This addresses the need for efficient and accurate models in NLP applications like question answering and dialogue, though it is incremental as it builds on existing parametric and retrieval-augmented approaches.
The paper tackled the problem of combining the computational efficiency of parametric models with the predictive accuracy of retrieval-augmented models for knowledge-intensive NLP tasks, resulting in improved accuracy (e.g., 25.8 to 44.3 EM on NQ) while maintaining high throughput (e.g., 1000 queries/s).
Access to external knowledge is essential for many natural language processing tasks, such as question answering and dialogue. Existing methods often rely on a parametric model that stores knowledge in its parameters, or use a retrieval-augmented model that has access to an external knowledge source. Parametric and retrieval-augmented models have complementary strengths in terms of computational efficiency and predictive accuracy. To combine the strength of both approaches, we propose the Efficient Memory-Augmented Transformer (EMAT) -- it encodes external knowledge into a key-value memory and exploits the fast maximum inner product search for memory querying. We also introduce pre-training tasks that allow EMAT to encode informative key-value representations, and to learn an implicit strategy to integrate multiple memory slots into the transformer. Experiments on various knowledge-intensive tasks such as question answering and dialogue datasets show that, simply augmenting parametric models (T5-base) using our method produces more accurate results (e.g., 25.8 -> 44.3 EM on NQ) while retaining a high throughput (e.g., 1000 queries/s on NQ). Compared to retrieval-augmented models, EMAT runs substantially faster across the board and produces more accurate results on WoW and ELI5. Our code and datasets are available at https://github. com/uclnlp/EMAT.