mlsanityData to 14 Oct 2026

Down and Across: Introducing Crossword-Solving as a New NLP Benchmark

arXiv 2205.10442 ↗First posted 20 May 2022cs.CL cs.AIcited by 4

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

Solving crossword puzzles requires diverse reasoning capabilities, access to a vast amount of knowledge about language and the world, and the ability to satisfy the constraints imposed by the structure of the puzzle. In this work, we introduce solving crossword puzzles as a new natural language understanding task. We release the specification of a corpus of crossword puzzles collected from the New York Times daily crossword spanning 25 years and comprised of a total of around nine thousand puzzles. These puzzles include a diverse set of clues: historic, factual, word meaning, synonyms/antonyms, fill-in-the-blank, abbreviations, prefixes/suffixes, wordplay, and cross-lingual, as well as clues that depend on the answers to other clues. We separately release the clue-answer pairs from these puzzles as an open-domain question answering dataset containing over half a million unique clue-answer pairs. For the question answering task, our baselines include several sequence-to-sequence and retrieval-based generative models. We also introduce a non-parametric constraint satisfaction baseline for solving the entire crossword puzzle. Finally, we propose an evaluation framework which consists of several complementary performance metrics.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 28 of its 42 references that are in the map. First posted on arXiv 20 May 2022.

Builds on

28 of 42 references.

References · 28

  1. 2017Attention Is All You Needcited by 60,361
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  4. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  5. 2019BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehensioncited by 5,893
  6. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,148
  7. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  8. 2020Dense Passage Retrieval for Open-Domain Question Answeringcited by 3,824
  9. 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,754
  10. 2020On Faithfulness and Factuality in Abstractive Summarizationcited by 900
  11. 2019Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inferencecited by 876
  12. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 722
16 more
  1. 2019Latent Retrieval for Weakly Supervised Open Domain Question Answeringcited by 695
  2. 2021Retrieval Augmentation Reduces Hallucination in Conversationcited by 692
  3. 2019Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailmentcited by 671
  4. 2019ELI5: Long Form Question Answeringcited by 575
  5. 2019Probing Neural Network Comprehension of Natural Language Argumentscited by 214
  6. 2018ReCoRD: Bridging the Gap between Human and Machine Commonsense Reading Comprehensioncited by 200
  7. 2015Large-scale Simple Question Answering with Memory Networkscited by 127
  8. 2021QA Dataset Explosion: A Taxonomy of NLP Resources for Question Answering and Reading Comprehensioncited by 91
  9. 2019Learning and Evaluating General Linguistic Intelligencecited by 79
  10. 2017Recurrent Relational Networkscited by 72
  11. 2020CharBERT: Character-aware Pre-trained Language Modelcited by 63
  12. 2019Assessing the Benchmarking Capacity of Machine Reading Comprehension Datasetscited by 46
  13. 2019What does BERT Learn from Multiple-Choice Reading Comprehension Datasets?cited by 30
  14. 2014Dr.Fill: Crosswords and an Implemented Solver for Singly Weighted CSPscited by 8
  15. 2019ORB: An Open Reading Benchmark for Comprehensive Evaluation of Machine Reading Comprehensioncited by 5
  16. 2021Reinforcement Learning For Constraint Satisfaction Game Agents (15-Puzzle, Minesweeper, 2048, and Sudoku)cited by 1

Built on by

Cited by year

Citing papers per year: 2024: 3; 2025: 1.

4 papers.

  1. 2024Puzzle Solving using Reasoning of Large Language Models: A Surveycited by 45
  2. 2025CrossWordBench: Evaluating the Reasoning Capabilities of LLMs and LVLMs with Controllable Puzzle Generationcited by 10
  3. 2024Language Models are Crossword Solverscited by 5
  4. 2024AcrosticSleuth: Probabilistic Identification and Ranking of Acrostics in Multilingual Corporacited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2023NLPBench: Evaluating Large Language Models on Solving NLP Problems
  2. 2025EnigmaEval: A Benchmark of Long Multimodal Reasoning Challenges
  3. 2025Cross-Question Method Reuse in Large Language Models: From Word-Level Prediction to Rational Logical-Layer Reasoning
  4. 2023Solving and Generating NPR Sunday Puzzles with Large Language Models
  5. 2026Breaking the Evaluation Paradox: Evaluating High-Entropy Search with Computationally Irreducible Constraints
  6. 2026From Reasoning Depth to Reasoning Breadth: Evaluating Multi-Point Associative Reasoning in Large Language Models
  7. 2026The Riddle Riddle: Testing Flexible Reasoning in Large Language Models and Humans
  8. 2024Navigating the Labyrinth: Evaluating LLMs' Ability to Reason About Search Problems