mlsanityData to 7 Oct 2026

LILO: Learning Interpretable Libraries by Compressing and Documenting Code

arXiv 2310.19791 ↗First posted 30 Oct 2023cs.CL cs.AI cs.LGcited by 46

Not on the map. Why?

Abstract (arXiv)

While large language models (LLMs) now excel at code generation, a key aspect of software development is the art of refactoring: consolidating code into libraries of reusable and readable programs. In this paper, we introduce LILO, a neurosymbolic framework that iteratively synthesizes, compresses, and documents code to build libraries tailored to particular problem domains. LILO combines LLM-guided program synthesis with recent algorithmic advances in automated refactoring from Stitch: a symbolic compression system that efficiently identifies optimal lambda abstractions across large code corpora. To make these abstractions interpretable, we introduce an auto-documentation (AutoDoc) procedure that infers natural language names and docstrings based on contextual examples of usage. In addition to improving human readability, we find that AutoDoc boosts performance by helping LILO's synthesizer to interpret and deploy learned abstractions. We evaluate LILO on three inductive program synthesis benchmarks for string editing, scene reasoning, and graphics composition. Compared to existing neural and symbolic methods - including the state-of-the-art library learning algorithm DreamCoder - LILO solves more complex tasks and learns richer libraries that are grounded in linguistic knowledge.

Key facts. Cited by 46 papers in the map (data through 7 Oct 2026). Builds on 44 of its 88 references that are in the map. First posted on arXiv 30 Oct 2023.

Builds on

44 of 88 references.

References · 44

  1. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  3. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,072
  4. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
  5. 2021Program Synthesis with Large Language Modelscited by 3,509
  6. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
  7. 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
  8. 2016CLEVR: A Diagnostic Dataset for Compositional Language and Elementary Visual Reasoningcited by 1,445
  9. 2022Competition-Level Code Generation with AlphaCodecited by 1,427
  10. 2021What Makes Good In-Context Examples for GPT-3?cited by 1,184
  11. 2021Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivitycited by 1,122
  12. 2022Large Language Models Are Human-Level Prompt Engineerscited by 1,020
32 more
  1. 2021Measuring Coding Challenge Competence With APPScited by 879
  2. 2022CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesiscited by 876
  3. 2022InCoder: A Generative Model for Code Infilling and Synthesiscited by 492
  4. 2015Neural Module Networkscited by 375
  5. 2023Chain-of-Verification Reduces Hallucination in Large Language Modelscited by 317
  6. 2018Neural-Symbolic VQA: Disentangling Reasoning from Vision and Language Understandingcited by 304
  7. 2022Memorizing Transformerscited by 219
  8. 2022Synchromesh: Reliable code generation from pre-trained language modelscited by 156
  9. 2016DeepCoder: Learning to Write Programscited by 145
  10. 2017Learning to Reason: End-to-End Module Networks for Visual Question Answeringcited by 125
  11. 2017RobustFill: Neural Program Learning under Noisy I/Ocited by 124
  12. 2021Skill Induction and Planning with Latent Languagecited by 99
  13. 2022Natural Language to Code Generation in Interactive Data Science Notebookscited by 86
  14. 2016Neuro-Symbolic Program Synthesiscited by 84
  15. 2017Learning with Latent Languagecited by 75
  16. 2022Language Models Can Teach Themselves to Program Bettercited by 75
  17. 2017Learning to Infer Graphics Programs from Hand-Drawn Imagescited by 73
  18. 2021DOBF: A Deobfuscation Pre-Training Objective for Programming Languagescited by 68
  19. 2022Top-Down Synthesis for Library Learningcited by 57
  20. 2021What Context Features Can Transformer Language Models Use?cited by 49
  21. 2019Learning to Infer Program Sketchescited by 44
  22. 2021Leveraging Language to Learn Program Abstractions and Search Heuristicscited by 44
  23. 2019Program Synthesis and Semantic Parsing with Learned Code Idiomscited by 34
  24. 2023The Larger They Are, the Harder They Fail: Language Models do not Recognize Identifier Swaps in Pythoncited by 29
  25. 20213DP3: 3D Scene Perception via Probabilistic Programmingcited by 26
  26. 2022babble: Learning Better Abstractions with E-Graphs and Anti-Unificationcited by 22
  27. 2018Beyond imitation: Zero-shot task transfer on robots by learning concepts as cognitive programscited by 21
  28. 2015Program Synthesis from Polymorphic Refinement Typescited by 20
  29. 2021Generating Adversarial Computer Programs using Optimized Obfuscationscited by 20
  30. 2014Mining Idioms from Source Codecited by 19
  31. 2017Agile Software Development Methods: Review and Analysiscited by 18
  32. 2021Multi-modal Program Inference: a Marriage of Pre-trainedLanguage Models and Component-based Synthesiscited by 12

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 15; 2025: 13; 2026: 18.

46 papers.

Sort:

By area:

  1. 2024Agent Workflow Memorycited by 342
  2. 2025Inducing Programmatic Skills for Agentic Taskscited by 103
  3. 2024WorldCoder, a Model-Based LLM Agent: Building World Models by Writing Code and Interacting with the Environmentcited by 99
  4. 2024Towards Guaranteed Safe AI: A Framework for Ensuring Robust and Reliable AI Systemscited by 94
  5. 2024Symbolic Regression with a Learned Concept Librarycited by 76
  6. 2024TroVE: Inducing Verifiable and Efficient Toolboxes for Solving Programmatic Taskscited by 70
  7. 2024What Are Tools Anyway? A Survey from the Language Model Perspectivecited by 53
  8. 2024ReGAL: Refactoring Programs to Discover Generalizable Abstractionscited by 32
  9. 2024Pix2Code: Learning to Compose Neural Visual Concepts as Programscited by 18
  10. 2026SkillDAG: Self-Evolving Typed Skill Graphs for LLM Skill Selection at Scalecited by 15
  11. 2025Modeling Open-World Cognition as On-Demand Synthesis of Probabilistic Modelscited by 14
  12. 2024Synthesizing Programmatic Reinforcement Learning Policies with Large Language Model Guided Searchcited by 13
34 more
  1. 2025Self-Steering Language Modelscited by 13
  2. 2024Loose LIPS Sink Ships: Asking Questions in Battleship with Language-Informed Program Samplingcited by 12
  3. 2026Executable World Models for ARC-AGI-3 in the Era of Coding Agentscited by 12
  4. 2025AI Agents as Universal Task Solverscited by 9
  5. 2025ShapeLib: Designing a library of programmatic 3D shape abstractions with Large Language Modelscited by 5
  6. 2024TurtleBench: A Visual Programming Benchmark in Turtle Geometrycited by 4
  7. 2025Refactoring Codebases through Library Designcited by 4
  8. 2026DreamProver: Evolving Transferable Lemma Libraries via a Wake-Sleep Theorem-Proving Agentcited by 4
  9. 2024Library Learning Doesn't: The Curious Case of the Single-Use "Library"cited by 3
  10. 2025GATE: Graph-based Adaptive Tool Evolution Across Diverse Taskscited by 3
  11. 2025LLM Library Learning Fails: A LEGO-Prover Case Studycited by 3
  12. 2025Characterizing Language Use in a Collaborative Situated Gamecited by 3
  13. 2026Do Coding Agents Need Executable World Models, Simplification, and Verification to Solve ARC-AGI-3?cited by 3
  14. 2024AbstractBeam: Enhancing Bottom-Up Program Synthesis using Library Learningcited by 2
  15. 2025Self-Evolving Visual Concept Library using Vision-Language Criticscited by 2
  16. 2025A Compute-Matched Re-Evaluation of TroVE on MATHcited by 2
  17. 2025Learning Interestingness in Automated Mathematical Theory Formationcited by 2
  18. 2025Transductive Visual Programming: Evolving Tool Libraries from Experience for Spatial Reasoningcited by 2
  19. 2026Artificial Intelligence and the Structure of Mathematicscited by 2
  20. 2026AllocBench: Measuring Online Tool Allocation Capability in LLM Agentscited by 2
  21. 2026ReaComp: Compiling LLM Reasoning into Symbolic Solvers for Efficient Program Synthesiscited by 1
  22. 2024Sketch-Plan-Generalize: Learning and Planning with Neuro-Symbolic Programmatic Representations for Inductive Spatial Conceptscited by 0
  23. 2024MathDSL: A Domain-Specific Language for Concise Mathematical Solutions Via Program Synthesiscited by 0
  24. 2026Prospective Compression in Human Abstraction Learningcited by 0
  25. 2026ShapeCodeBench: A Renewable Benchmark for Perception-to-Program Reconstruction of Synthetic Shape Scenescited by 0
  26. 2026Inducing Reasoning Primitives from Agent Tracescited by 0
  27. 2026When LLMs Develop Languages: Symbolic Communication for Efficient Multi-Agent Reasoningcited by 0
  28. 2026Atomic Units of X: The Compression Layer of Intelligencecited by 0
  29. 2026PACE: Primitive-Aware Code Evolution for Automated Algorithm Designcited by 0
  30. 2026SkillComposer: Learning Reusable Skills for Natural-Language Robot Programmingcited by 0
  31. 2026Super Library Agent: Joint Generation and Maintenance of Multiple Applications Beyond the Single Codebasecited by 0
  32. 2026REFACTOR-VLA: Unsupervised Library Learning of Typed Motor Programscited by 0
  33. 2026X-Tree: Tokenizing Reusable Experience for Efficient Agent Generalizationcited by 0
  34. 2026MatLoom: Layered Text-to-Material Generation in a Compact Program Spacecited by 0