The Natural Language Decathlon: Multitask Learning as Question Answering
This work addresses the need for general NLP models that can handle multiple tasks, moving beyond single-task focus, though it is incremental in its approach by building on existing multitask and question answering paradigms.
The paper tackles the problem of developing general NLP models by introducing the Natural Language Decathlon (decaNLP), a challenge spanning ten tasks cast as question answering, and presents the Multitask Question Answering Network (MQAN) that jointly learns all tasks without task-specific modules, achieving improvements in transfer learning, domain adaptation, and zero-shot capabilities, and state-of-the-art results on WikiSQL.
Deep learning has improved performance on many natural language processing (NLP) tasks individually. However, general NLP models cannot emerge within a paradigm that focuses on the particularities of a single metric, dataset, and task. We introduce the Natural Language Decathlon (decaNLP), a challenge that spans ten tasks: question answering, machine translation, summarization, natural language inference, sentiment analysis, semantic role labeling, zero-shot relation extraction, goal-oriented dialogue, semantic parsing, and commonsense pronoun resolution. We cast all tasks as question answering over a context. Furthermore, we present a new Multitask Question Answering Network (MQAN) jointly learns all tasks in decaNLP without any task-specific modules or parameters in the multitask setting. MQAN shows improvements in transfer learning for machine translation and named entity recognition, domain adaptation for sentiment analysis and natural language inference, and zero-shot capabilities for text classification. We demonstrate that the MQAN's multi-pointer-generator decoder is key to this success and performance further improves with an anti-curriculum training strategy. Though designed for decaNLP, MQAN also achieves state of the art results on the WikiSQL semantic parsing task in the single-task setting. We also release code for procuring and processing data, training and evaluating models, and reproducing all experiments for decaNLP.