mlsanityData to 7 Oct 2026

Compositional Generative Modeling: A Single Model is Not All You Need

arXiv 2402.01103 ↗First posted 2 Feb 2024cs.LG cs.AI cs.CVcited by 41

Not on the map. Why?

Abstract (arXiv)

Large monolithic generative models trained on massive amounts of data have become an increasingly dominant approach in AI research. In this paper, we argue that we should instead construct large generative systems by composing smaller generative models together. We show how such a compositional generative approach enables us to learn distributions in a more data-efficient manner, enabling generalization to parts of the data distribution unseen at training time. We further show how this enables us to program and construct new generative models for tasks completely unseen at training. Finally, we show that in many cases, we can discover separate compositional components from data.

Key facts. Cited by 41 papers in the map (data through 7 Oct 2026). Builds on 39 of its 51 references that are in the map. First posted on arXiv 2 Feb 2024.

Builds on

39 of 51 references.

References · 39

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  4. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  7. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,611
  8. 2022Training Compute-Optimal Large Language Modelscited by 2,927
  9. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
  10. 2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,023
  11. 2016Pixel Recurrent Neural Networkscited by 814
  12. 2022A Generalist Agentcited by 802
27 more
  1. 2023Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Modelscited by 695
  2. 2022Compositional Visual Generation with Composable Diffusion Modelscited by 619
  3. 2023Are Emergent Abilities of Large Language Models a Mirage?cited by 515
  4. 2022Is Conditional Generative Modeling all you need for Decision-Making?cited by 504
  5. 2023Faith and Fate: Limits of Transformers on Compositionalitycited by 496
  6. 2019Your Classifier is Secretly an Energy Based Model and You Should Treat it Like Onecited by 402
  7. 2023Reduce, Reuse, Recycle: Compositional Generation with Energy-Based Diffusion Models and MCMCcited by 216
  8. 2021Understanding the Capabilities, Limitations, and Societal Impact of Large Language Modelscited by 159
  9. 2020Improved Contrastive Divergence Training of Energy Based Modelscited by 127
  10. 2023Are Emergent Abilities in Large Language Models just In-Context Learning?cited by 109
  11. 2019On the Anatomy of MCMC-Based Maximum Likelihood Learning of Energy-Based Modelscited by 105
  12. 2023DiffCollage: Parallel Generation of Large Content with Diffusion Modelscited by 100
  13. 2021Oops I Took A Gradient: Scalable Sampling for Discrete Distributionscited by 92
  14. 2023Compositional Foundation Models for Hierarchical Planningcited by 92
  15. 2023Generative Skill Chaining: Long-Horizon Skill Planning with Diffusion Modelscited by 87
  16. 2021Unsupervised Learning of Compositional Energy Conceptscited by 71
  17. 2023Compositional Generalization from First Principlescited by 62
  18. 2023Additive Decoders for Latent Variables Identification and Cartesian-Product Extrapolationcited by 56
  19. 2023Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Modelscited by 45
  20. 2023Compositional Diffusion-Based Continuous Constraint Solverscited by 39
  21. 2023Unsupervised Compositional Concepts Discovery with Text-to-Image Generative Modelscited by 36
  22. 2022VAEL: Bridging Variational Autoencoders and Probabilistic Logic Programmingcited by 31
  23. 2023Probabilistic Adaptation of Text-to-Video Modelscited by 29
  24. 2022Composing Ensembles of Pre-trained Models via Iterative Consensuscited by 28
  25. 2023Compositional Sculpting of Iterative Generative Processescited by 26
  26. 2023Neurosymbolic Grounding for Compositional World Modelscited by 14
  27. 2023Inferring Relational Potentials in Interacting Systemscited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 13; 2025: 22; 2026: 6.

41 papers.

Sort:

By area:

  1. 2024How Far is Video Generation from World Model: A Physical Law Perspectivecited by 204
  2. 2025History-Guided Video Diffusioncited by 165
  3. 2025Feynman-Kac Correctors in Diffusion: Annealing, Guidance, and Product of Expertscited by 85
  4. 2024ConceptMix: A Compositional Image Generation Benchmark with Controllable Difficultycited by 60
  5. 2024Motion-Agent: A Conversational Framework for Human Motion Generation with LLMscited by 46
  6. 2025Generative Trajectory Stitching through Diffusion Compositioncited by 38
  7. 2024The Superposition of Diffusion Models Using the Ito Density Estimatorcited by 36
  8. 2024Deep Generative Models in Robotics: A Survey on Learning from Multimodal Demonstrationscited by 35
  9. 2024Can Models Learn Skill Composition from Examples?cited by 17
  10. 2024Agreement-Based Cascading for Efficient Inferencecited by 16
  11. 2024MeanCache: User-Centric Semantic Caching for LLM Web Servicescited by 14
  12. 2025Does Data Scaling Lead to Visual Compositional Generalization?cited by 13
29 more
  1. 2024Progressive Compositionality in Text-to-Image Generative Modelscited by 11
  2. 2025Multi-Modal Manipulation via Multi-Modal Policy Consensuscited by 9
  3. 2025Compose Your Policies! Improving Diffusion-based or Flow-based Robot Policies via Test-time Distribution-level Compositioncited by 8
  4. 2025Unconditional Priors Matter! Improving Conditional Generation of Fine-Tuned Diffusion Modelscited by 7
  5. 2025Advancing Semantic Caching for LLMs with Domain-Specific Embeddings and Synthetic Datacited by 7
  6. 2025Scaling can lead to compositional generalizationcited by 7
  7. 2025Product of Experts for Visual Generationcited by 6
  8. 2024Swing-by Dynamics in Concept Learning and Compositional Generalizationcited by 5
  9. 2025Modality-Composable Diffusion Policy via Inference-Time Distribution-level Compositioncited by 5
  10. 2025Learning Primitive Embodied World Models: Towards Scalable Robotic Learningcited by 5
  11. 2025Flexible Multitask Learning with Factorized Diffusion Policycited by 5
  12. 2025Compositional Scene Understanding through Inverse Generative Modelingcited by 3
  13. 2025Nonparametric Identification of Latent Conceptscited by 3
  14. 2026Compositional Visual Planning via Inference-Time Diffusion Scalingcited by 2
  15. 2025An Affective-Taxis Hypothesis for Alignment and Interpretabilitycited by 1
  16. 2025Intervening to Learn and Compose Causally Disentangled Representationscited by 1
  17. 2025Ctrl-VI: Controllable Video Synthesis via Variational Inferencecited by 1
  18. 2025Learning by Analogy: A Causal Framework for Composition Generalizationcited by 1
  19. 2026Meanings and Measurements: Multi-Agent Probabilistic Grounding for Vision-Language Navigationcited by 1
  20. 2026Long-Text-to-Image Generation via Compositional Prompt Decompositioncited by 1
  21. 2026A Blueprint for Equilibrium-Based Differentiable Continuous-Variable Thermodynamic Computingcited by 1
  22. 2024Let the Quantum Creep In: Designing Quantum Neural Network Models by Gradually Swapping Out Classical Componentscited by 0
  23. 2024Absorb & Escape: Overcoming Single Model Limitations in Generating Genomic Sequencescited by 0
  24. 2024ResidualDroppath: Enhancing Feature Reuse over Residual Connectionscited by 0
  25. 2025What Do You Need for Compositional Generalization in Diffusion Planning?cited by 0
  26. 2025CanvasMAR: Improving Masked Autoregressive Video Prediction With Canvascited by 0
  27. 2025DeMoGen: Towards Decompositional Human Motion Generation with Energy-Based Diffusion Modelscited by 0
  28. 2026Conflict-Aware Additive Guidance for Flow Models under Compositional Rewardscited by 0
  29. 2026Visual General Intelligence: A White Papercited by 0