# Ask Question First for Enhancing Lifelong Language Learning

arXiv: https://arxiv.org/abs/2208.08367 · HTML: https://mlsanity.com/paper/2208.08367/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 30 references that are in the map.
- First posted on arXiv 17 Aug 2022.

## Builds on (most cited first, 13 of 13)

- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [The E2E Dataset: New Challenges For End-to-End Generation](https://mlsanity.com/paper/1706.09254/) (2017)
- [The Natural Language Decathlon: Multitask Learning as Question Answering](https://mlsanity.com/paper/1806.08730/) (2018)
- [LAMOL: LAnguage MOdeling for Lifelong Language Learning](https://mlsanity.com/paper/1909.03329/) (2019)
- [Learning to Generate Reviews and Discovering Sentiment](https://mlsanity.com/paper/1704.01444/) (2017)
- [Semantically Conditioned LSTM-based Natural Language Generation for Spoken Dialogue Systems](https://mlsanity.com/paper/1508.01745/) (2015)
- [Invariant Rationalization](https://mlsanity.com/paper/2003.09772/) (2020)
- [Continual Learning for Text Classification with Information Disentanglement Based Regularization](https://mlsanity.com/paper/2104.05489/) (2021)
- [Continual Learning in Task-Oriented Dialogue Systems](https://mlsanity.com/paper/2012.15504/) (2020)
- [Continual Learning for Natural Language Generation in Task-oriented Dialog Systems](https://mlsanity.com/paper/2010.00910/) (2020)
- [Lifelong Language Knowledge Distillation](https://mlsanity.com/paper/2010.02123/) (2020)
- [Continual Learning for Sentence Representations Using Conceptors](https://mlsanity.com/paper/1904.09187/) (2019)
- [Reminding the Incremental Language Model via Data-Free Self-Distillation](https://mlsanity.com/paper/2110.08745/) (2021)

## Abstract (arXiv)

Lifelong language learning aims to stream learning NLP tasks while retaining knowledge of previous tasks. Previous works based on the language model and following data-free constraint approaches have explored formatting all data as "begin token (textitB) + context (textitC) + question (textitQ) + answer (textitA)" for different tasks. However, they still suffer from catastrophic forgetting and are exacerbated when the previous task's pseudo data is insufficient for the following reasons: (1) The model has difficulty generating task-corresponding pseudo data, and (2) textitA is prone to error when textitA and textitC are separated by textitQ because the information of the textitC is diminished before generating textitA. Therefore, we propose the Ask Question First and Replay Question (AQF-RQ), including a novel data format "textitBQCA" and a new training task to train pseudo questions of previous tasks. Experimental results demonstrate that AQF-RQ makes it easier for the model to generate more pseudo data that match corresponding tasks, and is more robust to both sufficient and insufficient pseudo-data when the task boundary is both clear and unclear. AQF-RQ can achieve only 0.36% lower performance than multi-task learning.