mlsanityData to 14 Oct 2026

Is a Question Decomposition Unit All We Need?

arXiv 2205.12538 ↗First posted 25 May 2022cs.CL cs.AI cs.HCcited by 37

Not on the map. Why?

Abstract (arXiv)

Large Language Models (LMs) have achieved state-of-the-art performance on many Natural Language Processing (NLP) benchmarks. With the growing number of new benchmarks, we build bigger and more complex LMs. However, building new LMs may not be an ideal option owing to the cost, time and environmental impact associated with it. We explore an alternative route: can we modify data by expressing it in terms of the model's strengths, so that a question becomes easier for models to answer? We investigate if humans can decompose a hard question into a set of simpler questions that are relatively easier for models to solve. We analyze a range of datasets involving various forms of reasoning and find that it is indeed possible to significantly improve model performance (24% for GPT3 and 29% for RoBERTa-SQuAD along with a symbolic calculator) via decomposition. Our approach provides a viable option to involve people in NLP research in a meaningful way. Our findings indicate that Human-in-the-loop Question Decomposition (HQD) can potentially provide an alternate path to building large LMs. Code and data is available at https://github.com/Pruthvi98/QuestionDecomposition

Key facts. Cited by 37 papers in the map (data through 14 Oct 2026). Builds on 31 of its 46 references that are in the map. First posted on arXiv 25 May 2022.

Builds on

31 of 46 references.

References · 31

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
  4. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  5. 2019BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehensioncited by 5,893
  6. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,780
  7. 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,754
  8. 2021Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processingcited by 2,234
  9. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,183
  10. 2018Know What You Don't Know: Unanswerable Questions for SQuADcited by 1,554
  11. 2021Multitask Prompted Training Enables Zero-Shot Task Generalizationcited by 1,470
  12. 2019DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphscited by 903
19 more
  1. 2021Are NLP Models really able to Solve Simple Math Word Problems?cited by 878
  2. 2021Did Aristotle Use a Laptop? A Question Answering Benchmark with Implicit Reasoning Strategiescited by 743
  3. 2021Cross-Task Generalization via Natural Language Crowdsourcing Instructionscited by 699
  4. 2019MathQA: Towards Interpretable Math Word Problem Solving with Operation-Based Formalismscited by 662
  5. 2018The Web as a Knowledge-base for Answering Complex Questionscited by 410
  6. 2019QASC: A Dataset for Question Answering via Sentence Compositioncited by 252
  7. 2019Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representationcited by 210
  8. 2021Reframing Instructional Prompts to GPTk's Languagecited by 176
  9. 2020Break It Down: A Question Understanding Benchmarkcited by 131
  10. 2019Multi-hop Reading Comprehension through Question Decomposition and Rescoringcited by 119
  11. 2020Unsupervised Question Decomposition for Question Answeringcited by 118
  12. 2022NumGLUE: A Suite of Fundamental yet Challenging Mathematical Reasoning Taskscited by 112
  13. 2020Text Modular Networks: Learning to Decompose Tasks in the Language of Existing Modelscited by 65
  14. 2022Less is More: Summary of Long Instructions is Better for Program Synthesiscited by 17
  15. 2022How Many Data Samples is an Additional Instruction Worth?cited by 16
  16. 2018Neural Compositional Denotational Semantics for Question Answeringcited by 11
  17. 2022Choose Your QA Model Wisely: A Systematic Study of Generative and Extractive Readers for Question Answeringcited by 11
  18. 2022Calibrating Trust of Multi-Hop Question Answering Systems with Decompositional Probescited by 9
  19. 2021A Simple Approach to Jointly Rank Passages and Select Relevant Sentences in the OBQA Contextcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 4; 2023: 13; 2024: 12; 2025: 5; 2026: 3.

37 papers.

Sort:

By area:

  1. 2022Measuring and Narrowing the Compositionality Gap in Language Modelscited by 1,075
  2. 2023Augmented Language Models: a Surveycited by 403
  3. 2023ToolQA: A Dataset for LLM Question Answering with External Toolscited by 274
  4. 2023An LLM Compiler for Parallel Function Callingcited by 179
  5. 2023Take a Step Back: Evoking Reasoning via Abstraction in Large Language Modelscited by 159
  6. 2022Ask Me Anything: A simple strategy for prompting language modelscited by 155
  7. 2024Self-Discover: Large Language Models Self-Compose Reasoning Structurescited by 110
  8. 2023Question Decomposition Improves the Faithfulness of Model-Generated Reasoningcited by 108
  9. 2024LogicBench: Towards Systematic Evaluation of Logical Reasoning Ability of Large Language Modelscited by 68
  10. 2023Natural Language Reasoning, A Surveycited by 61
  11. 2025ChemAgent: Self-updating Library in Large Language Models Improves Chemical Reasoningcited by 41
  12. 2024MoreHopQA: More Than Multi-hop Reasoningcited by 37
25 more
  1. 2023InstructABSA: Instruction Learning for Aspect Based Sentiment Analysiscited by 23
  2. 2023Instruction Tuned Models are Quick Learnerscited by 21
  3. 2025Minions: Cost-efficient Collaboration Between On-device and Cloud Language Modelscited by 20
  4. 2025Collab-RAG: Boosting Retrieval-Augmented Generation for Complex Question Answering via White-Box and Black-Box LLM Collaborationcited by 20
  5. 2022HELP ME THINK: A Simple Prompting Strategy for Non-experts to Create Customized Content with Modelscited by 17
  6. 2023InstructExcel: A Benchmark for Natural Language Instruction in Excelcited by 15
  7. 2023Exploring Question Decomposition for Zero-Shot VQAcited by 12
  8. 2023Large Language Model Programscited by 11
  9. 2024Plan*RAG: Efficient Test-Time Planning for Retrieval Augmented Generationcited by 11
  10. 2024SQLucid: Grounding Natural Language Database Queries with Interactive Explanationscited by 10
  11. 2024Skin-in-the-Game: Decision Making via Multi-Stakeholder Alignment in LLMscited by 8
  12. 2024Visual Question Decomposition on Multimodal Large Language Modelscited by 7
  13. 2025Fast-Slow-Thinking: Complex Task Solving with Large Language Modelscited by 7
  14. 2023DialCoT Meets PPO: Decomposing and Exploring Reasoning Paths in Smaller Language Modelscited by 6
  15. 2024FinQAPT: Empowering Financial Decisions with End-to-End LLM-driven Question Answering Pipelinecited by 6
  16. 2025PLAN-TUNING: Post-Training Language Models to Learn Step-by-Step Planning for Complex Problem Solvingcited by 6
  17. 2023In-Context Ability Transfer for Question Decomposition in Complex QAcited by 3
  18. 2024GenSco: Can Question Decomposition based Passage Alignment improve Question Answering?cited by 3
  19. 2026Quantifying the Statistical Effect of Rubric Modifications on Human-Autorater Agreementcited by 3
  20. 2022When Do Decompositions Help for Machine Reading?cited by 2
  21. 2026Say It My Way: Exploring Control in Conversational Visual Question Answering with Blind Userscited by 2
  22. 2024Review-Then-Refine: A Dynamic Framework for Multi-Hop Question Answering with Temporal Adaptabilitycited by 1
  23. 2024NeSy is alive and well: A LLM-driven symbolic approach for better code comment data generation and classificationcited by 0
  24. 2024FS-RAG: A Frame Semantics Based Approach for Improved Factual Accuracy in Large Language Modelscited by 0
  25. 2026A Navigational Approach for Comprehensive RAG via Traversal over Proposition Graphscited by 0