# Can Open-Domain QA Reader Utilize External Knowledge Efficiently like Humans?

arXiv: https://arxiv.org/abs/2211.12707 · HTML: https://mlsanity.com/paper/2211.12707/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 48 of its 55 references that are in the map.
- First posted on arXiv 23 Nov 2022.

## Builds on (most cited first, 30 of 48)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [REALM: Retrieval-Augmented Language Model Pre-Training](https://mlsanity.com/paper/2002.08909/) (2020)
- [BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-models](https://mlsanity.com/paper/2106.10199/) (2021)
- [Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering](https://mlsanity.com/paper/2007.01282/) (2020)
- [ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT](https://mlsanity.com/paper/2004.12832/) (2020)
- [TinyBERT: Distilling BERT for Natural Language Understanding](https://mlsanity.com/paper/1909.10351/) (2019)
- [Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inference](https://mlsanity.com/paper/2001.07676/) (2020)
- [Reading Wikipedia to Answer Open-Domain Questions](https://mlsanity.com/paper/1704.00051/) (2017)
- [How Much Knowledge Can You Pack Into the Parameters of a Language Model?](https://mlsanity.com/paper/2002.08910/) (2020)
- [Improved Knowledge Distillation via Teacher Assistant](https://mlsanity.com/paper/1902.03393/) (2019)
- [Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT](https://mlsanity.com/paper/1909.05840/) (2019)
- [Parameter-Efficient Transfer Learning with Diff Pruning](https://mlsanity.com/paper/2012.07463/) (2020)
- [DeeBERT: Dynamic Early Exiting for Accelerating BERT Inference](https://mlsanity.com/paper/2004.12993/) (2020)
- [DynaBERT: Dynamic BERT with Adaptive Width and Depth](https://mlsanity.com/paper/2004.04037/) (2020)
- [Selective Question Answering under Domain Shift](https://mlsanity.com/paper/2006.09462/) (2020)
- [Structured Pruning of Large Language Models](https://mlsanity.com/paper/1910.04732/) (2019)
- [PAQ: 65 Million Probably-Asked Questions and What You Can Do With Them](https://mlsanity.com/paper/2102.07033/) (2021)
- [TernaryBERT: Distillation-aware Ultra-low Bit BERT](https://mlsanity.com/paper/2009.12812/) (2020)
- [Learning Dense Representations of Phrases at Scale](https://mlsanity.com/paper/2012.12624/) (2020)
- [BAM! Born-Again Multi-Task Networks for Natural Language Understanding](https://mlsanity.com/paper/1907.04829/) (2019)
- [Compression of Generative Pre-trained Language Models via Quantization](https://mlsanity.com/paper/2203.10705/) (2022)
- [A Discrete Hard EM Approach for Weakly Supervised Question Answering](https://mlsanity.com/paper/1909.04849/) (2019)
- [Length-Adaptive Transformer: Train Once with Length Drop, Use Anytime with Search](https://mlsanity.com/paper/2010.07003/) (2020)
- [Towards Zero-Label Language Learning](https://mlsanity.com/paper/2109.09193/) (2021)
- [Investigating Selective Prediction Approaches Across Several Tasks in IID, OOD, and Adversarial Settings](https://mlsanity.com/paper/2203.00211/) (2022)

## Built on by (most cited first, 10 of 10)

- [Active Retrieval Augmented Generation](https://mlsanity.com/paper/2305.06983/) (2023)
- [CoLT5: Faster Long-Range Transformers with Conditional Computation](https://mlsanity.com/paper/2303.09752/) (2023)
- [The Art of Defending: A Systematic Evaluation and Analysis of LLM Defense Strategies on Safety and Over-Defensiveness](https://mlsanity.com/paper/2401.00287/) (2023)
- [FiDO: Fusion-in-Decoder optimized for stronger performance and faster inference](https://mlsanity.com/paper/2212.08153/) (2022)
- [Post-Abstention: Towards Reliably Re-Attempting the Abstained Instances in QA](https://mlsanity.com/paper/2305.01812/) (2023)
- [Pre-computed memory or on-the-fly encoding? A hybrid approach to retrieval augmentation makes the most of your compute](https://mlsanity.com/paper/2301.10448/) (2023)
- [Learning to Insert [PAUSE] Tokens for Better Reasoning](https://mlsanity.com/paper/2506.03616/) (2025)
- [GLIMMER: generalized late-interaction memory reranker](https://mlsanity.com/paper/2306.10231/) (2023)
- [Attention-Aligned Reasoning for Large Language Models](https://mlsanity.com/paper/2510.03223/) (2025)
- [A Unified Evaluation Framework for Novelty Detection and Accommodation in NLP with an Instantiation in Authorship Attribution](https://mlsanity.com/paper/2305.05079/) (2023)

## Abstract (arXiv)

Recent state-of-the-art open-domain QA models are typically based on a two stage retriever-reader approach in which the retriever first finds the relevant knowledge/passages and the reader then leverages that to predict the answer. Prior work has shown that the performance of the reader usually tends to improve with the increase in the number of these passages. Thus, state-of-the-art models use a large number of passages (e.g. 100) for inference. While the reader in this approach achieves high prediction performance, its inference is computationally very expensive. We humans, on the other hand, use a more efficient strategy while answering: firstly, if we can confidently answer the question using our already acquired knowledge then we do not even use the external knowledge, and in the case when we do require external knowledge, we don't read the entire knowledge at once, instead, we only read that much knowledge that is sufficient to find the answer. Motivated by this procedure, we ask a research question "Can the open-domain QA reader utilize external knowledge efficiently like humans without sacrificing the prediction performance?" Driven by this question, we explore an approach that utilizes both 'closed-book' (leveraging knowledge already present in the model parameters) and 'open-book' inference (leveraging external knowledge). Furthermore, instead of using a large fixed number of passages for open-book inference, we dynamically read the external knowledge in multiple 'knowledge iterations'. Through comprehensive experiments on NQ and TriviaQA datasets, we demonstrate that this dynamic reading approach improves both the 'inference efficiency' and the 'prediction accuracy' of the reader. Comparing with the FiD reader, this approach matches its accuracy by utilizing just 18.32% of its reader inference cost and also outperforms it by achieving up to 55.10% accuracy on NQ Open.