# NarrativeXL: A Large-scale Dataset For Long-Term Memory Models

arXiv: https://arxiv.org/abs/2305.13877 · HTML: https://mlsanity.com/paper/2305.13877/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 27 references that are in the map.
- First posted on arXiv 23 May 2023.

## Builds on (most cited first, 22 of 22)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [Longformer: The Long-Document Transformer](https://mlsanity.com/paper/2004.05150/) (2020)
- [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://mlsanity.com/paper/1901.02860/) (2019)
- [ChatGPT Outperforms Crowd-Workers for Text-Annotation Tasks](https://mlsanity.com/paper/2303.15056/) (2023)
- [Efficient Transformers: A Survey](https://mlsanity.com/paper/2009.06732/) (2020)
- [Compressive Transformers for Long-Range Sequence Modelling](https://mlsanity.com/paper/1911.05507/) (2019)
- [Long Range Arena: A Benchmark for Efficient Transformers](https://mlsanity.com/paper/2011.04006/) (2020)
- [The NarrativeQA Reading Comprehension Challenge](https://mlsanity.com/paper/1712.07040/) (2017)
- [A Discourse-Aware Attention Model for Abstractive Summarization of Long Documents](https://mlsanity.com/paper/1804.05685/) (2018)
- [QMSum: A New Benchmark for Query-based Multi-domain Meeting Summarization](https://mlsanity.com/paper/2104.05938/) (2021)
- [Towards AI-Complete Question Answering: A Set of Prerequisite Toy Tasks](https://mlsanity.com/paper/1502.05698/) (2015)
- [Modifying Memories in Transformer Models](https://mlsanity.com/paper/2012.00363/) (2020)
- [QuALITY: Question Answering with Long Input Texts, Yes!](https://mlsanity.com/paper/2112.08608/) (2021)
- [Exploring Length Generalization in Large Language Models](https://mlsanity.com/paper/2207.04901/) (2022)
- [Recurrent Memory Transformer](https://mlsanity.com/paper/2207.06881/) (2022)
- [BookSum: A Collection of Datasets for Long-form Narrative Summarization](https://mlsanity.com/paper/2105.08209/) (2021)
- [SCROLLS: Standardized CompaRison Over Long Language Sequences](https://mlsanity.com/paper/2201.03533/) (2022)
- [Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4](https://mlsanity.com/paper/2305.00118/) (2023)
- [What Gives the Answer Away? Question Answering Bias Analysis on Video QA Datasets](https://mlsanity.com/paper/2007.03626/) (2020)
- [Updater-Extractor Architecture for Inductive World State Representations](https://mlsanity.com/paper/2104.05500/) (2021)

## Built on by (most cited first, 2 of 2)

- [NovelQA: Benchmarking Question Answering on Documents Exceeding 200K Tokens](https://mlsanity.com/paper/2403.12766/) (2024)
- [Collaborative QA using Interacting LLMs. Impact of Network Structure, Node Capability and Distributed Data](https://mlsanity.com/paper/2511.14098/) (2025)

## Abstract (arXiv)

We propose a new large-scale (nearly a million questions) ultra-long-context (more than 50,000 words average document length) reading comprehension dataset. Using GPT 3.5, we summarized each scene in 1,500 hand-curated fiction books from Project Gutenberg, which resulted in approximately 150 scene-level summaries per book. After that, we created a number of reading comprehension questions based on these summaries, including three types of multiple-choice scene recognition questions, as well as free-form narrative reconstruction questions. With 990,595 total questions, our dataset is an order of magnitude larger than the closest alternatives. Crucially, most questions have a known ``retention demand'', indicating how long-term of a memory is needed to answer them, which should aid long-term memory performance evaluation. We validate our data in four small-scale experiments: one with human labelers, and three with existing language models. We show that our questions 1) adequately represent the source material 2) can be used to diagnose a model's memory capacity 3) are not trivial for modern language models even when the memory demand does not exceed those models' context lengths. Lastly, we provide our code which can be used to further expand the dataset with minimal human labor.