mlsanityData to 7 Oct 2026

Iterated Decomposition: Improving Science Q&A by Supervising Reasoning Processes

arXiv 2301.01751 ↗First posted 4 Jan 2023cs.CL cs.AI cs.HCcited by 15

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

Language models (LMs) can perform complex reasoning either end-to-end, with hidden latent state, or compositionally, with transparent intermediate state. Composition offers benefits for interpretability and safety, but may need workflow support and infrastructure to remain competitive. We describe iterated decomposition, a human-in-the-loop workflow for developing and refining compositional LM programs. We improve the performance of compositions by zooming in on failing components and refining them through decomposition, additional context, chain of thought, etc. To support this workflow, we develop ICE, an open-source tool for visualizing the execution traces of LM programs. We apply iterated decomposition to three real-world tasks and improve the accuracy of LM programs over less compositional baselines: describing the placebo used in a randomized controlled trial (25% to 65%), evaluating participant adherence to a medical intervention (53% to 70%), and answering NLP questions on the Qasper dataset (38% to 69%). These applications serve as case studies for a workflow that, if automated, could keep ML systems interpretable and safe even as they scale to increasingly complex tasks.

Key facts. Cited by 15 papers in the map (data through 7 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 43 of its 55 references that are in the map. First posted on arXiv 4 Jan 2023.

Builds on

43 of 55 references.

Closest ancestors · 7

Same direction, newest first

  1. 2022LAMBADA: Backward Chaining for Automated Reasoning in Natural Language20 Dec
  2. 2022PAL: Program-aided Language Models18 Nov
  3. 2022Measuring and Narrowing the Compositionality Gap in Language Models7 Oct
  4. 2022ReAct: Synergizing Reasoning and Acting in Language Models6 Oct
  5. 2022Faithful Reasoning Using Large Language Models30 Aug
  6. 2022Selection-Inference: Exploiting Large Language Models for Interpretable Logical Reasoning19 May
  7. 2022FaiRR: Faithful and Robust Deductive Reasoning over Natural Language19 Mar

Other references · 36

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2017Deep reinforcement learning from human preferencescited by 3,846
  3. 2020Learning to summarize from human feedbackcited by 2,562
  4. 2019Fine-Tuning Language Models from Human Preferencescited by 2,124
  5. 2021WebGPT: Browser-assisted question-answering with human feedbackcited by 1,722
  6. 2016Concrete Problems in AI Safetycited by 1,582
  7. 2022Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questionscited by 911
  8. 2022STaR: Bootstrapping Reasoning With Reasoningcited by 895
  9. 2022Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learningcited by 886
  10. 2022Solving math word problems with process- and outcome-based feedbackcited by 871
  11. 2022Decomposed Prompting: A Modular Approach for Solving Complex Taskscited by 554
  12. 2021A Dataset of Information-Seeking Questions and Answers Anchored in Research Paperscited by 475
24 more
  1. 2022Self-critiquing models for assisting human evaluatorscited by 372
  2. 2022Demonstrate-Search-Predict: Composing retrieval and language models for knowledge-intensive NLPcited by 316
  3. 2021Recursively Summarizing Books with Human Feedbackcited by 287
  4. 2020Pretrained Transformers for Text Ranking: BERT and Beyondcited by 266
  5. 2021AI Chains: Transparent and Controllable Human-AI Interaction by Chaining Large Language Model Promptscited by 265
  6. 2022The Alignment Problem from a Deep Learning Perspectivecited by 246
  7. 2022Re3: Generating Longer Stories With Recursive Reprompting and Revisioncited by 193
  8. 2022Maieutic Prompting: Logically Consistent Reasoning with Recursive Explanationscited by 159
  9. 2022PromptChainer: Chaining Large Language Model Prompts through Visual Programmingcited by 134
  10. 2022Language Model Cascadescited by 104
  11. 2021Decontextualization: Making Sentences Stand-Alonecited by 101
  12. 2020NILE : Natural Language Inference with Faithful Natural Language Explanationscited by 96
  13. 2022Compositional Semantic Parsing with Large Language Modelscited by 95
  14. 2022Successive Prompting for Decomposing Complex Questionscited by 94
  15. 2022Sub-Task Decomposition Enables Learning in Sequence to Sequence Taskscited by 41
  16. 2021Decomposing Complex Questions Makes Multi-Hop QA Easier and More Interpretablecited by 35
  17. 2022Natural Language Deduction through Search over Statement Compositionscited by 35
  18. 2022ThinkSum: Probabilistic reasoning over sets using large language modelscited by 23
  19. 2022Honest Students from Untrusted Teachers: Learning an Interpretable Question-Answering Pipeline from a Pretrained Language Modelcited by 20
  20. 2022Learning to Decompose: Hypothetical Question Decomposition Based on Comparable Textscited by 18
  21. 2022Summarization Programs: Interpretable Abstractive Summarization with Neural Modular Treescited by 14
  22. 2022Calibrating Trust of Multi-Hop Question Answering Systems with Decompositional Probescited by 9
  23. 2022Complex Reading Comprehension Through Question Decompositioncited by 7
  24. 2022Locate Then Ask: Interpretable Stepwise Reasoning for Multi-hop Question Answeringcited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 6; 2024: 2; 2025: 5; 2026: 2.

15 papers.

Sort:

By area:

  1. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,081
  2. 2023Question Decomposition Improves the Faithfulness of Model-Generated Reasoningcited by 107
  3. 2023Large Language Models as Tax Attorneys: A Case Study in Legal Capabilities Emergencecited by 42
  4. 2023Designing LLM Chains by Adapting Techniques from Crowdsourcing Workflowscited by 17
  5. 2024SMARTCAL: An Approach to Self-Aware Tool-Use Evaluation and Calibrationcited by 13
  6. 2023Large Language Model Programscited by 11
  7. 2024Enhancing Robustness in Large Language Models: Prompting for Mitigating the Impact of Irrelevant Informationcited by 8
  8. 2023KwaiYiiMath: Technical Reportcited by 4
  9. 2026DTCRS: Dynamic Tree Construction for Recursive Summarizationcited by 4
  10. 2025AdaptBot: Combining LLM with Knowledge Graphs and Human Input for Generic-to-Specific Task Decomposition and Knowledge Refinementcited by 3
  11. 2026How to Steer Your Multi-Agent System: Human-LLM Collaborative Planningcited by 2
  12. 2025DeKeyNLU: Enhancing Natural Language to SQL Generation through Task Decomposition and Keyword Extractioncited by 1
3 more
  1. 2025Robust Native Language Identification through Agentic Decompositioncited by 1
  2. 2025Dynamic Template Selection for Output Token Generation Optimization: MLP-Based and Transformer Approachescited by 0
  3. 2025A Flexible Multi-Agent LLM-Human Framework for Fast Human Validated Tool Buildingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025The Thinking Spectrum: An Empirical Study of Tunable Reasoning in LLMs through Model Merging
  2. 2024A NotSo Simple Way to Beat Simple Bench
  3. 2025Can Language Models Falsify? Evaluating Algorithmic Reasoning with Counterexample Creation
  4. 2025LLMSR@XLLM25: An Empirical Study of LLM for Structural Reasoning
  5. 2024Hierarchical Deconstruction of LLM Reasoning: A Graph-Based Framework for Analyzing Knowledge Utilization
  6. 2024textttLM^texttt2: A Simple Society of Language Models Solves Complex Reasoning
  7. 2026ReasoningFlow: Discourse Structures for Understanding LLM Reasoning Traces
  8. 2025Dissecting Logical Reasoning in LLMs: A Fine-Grained Evaluation and Supervision Study