mlsanityData to 14 Oct 2026

Model Collapse in the Self-Consuming Chain of Diffusion Finetuning: A Novel Perspective from Quantitative Trait Modeling

arXiv 2407.17493 ↗First posted 4 Jul 2024cs.CV cs.AIcited by 12

On the map: Synthetic data and self-improvement for LLMs → Model collapse from synthetic data

Abstract (arXiv)

Model collapse, the severe degradation of generative models when iteratively trained on their own outputs, has gained significant attention in recent years. This paper examines Chain of Diffusion, where a pretrained text-to-image diffusion model is finetuned on its own generated images. We demonstrate that severe image quality degradation was universal and identify CFG scale as the key factor impacting this model collapse. Drawing on an analogy between the Chain of Diffusion and biological evolution, we then introduce a novel theoretical analysis based on quantitative trait modeling from statistical genetics. Our theoretical analysis aligns with empirical observations of the generated images in the Chain of Diffusion. Finally, we propose Reusable Diffusion Finetuning (ReDiFine), a simple yet effective strategy inspired by genetic mutations. It operates robustly across various scenarios without requiring any hyperparameter tuning, making it a plug-and-play solution for reusable image generation.

Key facts. Cited by 12 papers in the map (data through 14 Oct 2026). Placed in Model collapse from synthetic data, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 35 of its 64 references that are in the map. First posted on arXiv 4 Jul 2024.

Builds on

35 of 64 references.

Closest ancestors · 5

Same direction, newest first

  1. 2024Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification11 Jun
  2. 2024Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data1 Apr
  3. 2024Model Collapse Demystified: The Case of Regression12 Feb
  4. 2024A Tale of Tails: Model Collapse as a Change of Scaling Laws10 Feb
  5. 2023The Curse of Recursion: Training on Generated Data Makes Models Forget27 May

Other references · 30

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  3. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  4. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  5. 2022Classifier-Free Diffusion Guidancecited by 5,847
  6. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
  7. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  8. 2014Deep Learning Face Attributes in the Wildcited by 3,693
  9. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 944
  10. 2022eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoiserscited by 931
  11. 2022Large Language Models Can Self-Improvecited by 664
  12. 2024Guiding a Diffusion Model with a Bad Version of Itselfcited by 299
18 more
  1. 2022On the detection of synthetic images generated by diffusion modelscited by 283
  2. 2024Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Modelscited by 254
  3. 2023Self-Consuming Generative Models Go MADcited by 250
  4. 2023Exposing flaws of generative model evaluation metrics and their unfair treatment of diffusion modelscited by 205
  5. 2023CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Samplingcited by 129
  6. 2023Intriguing properties of synthetic images: from generative adversarial networks to diffusion modelscited by 123
  7. 2023On the Stability of Iterative Retraining of Generative Models on their own Datacited by 100
  8. 2023The Curious Decline of Linguistic Diversity: Training Language Models on Synthetic Textcited by 100
  9. 2023Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loopcited by 79
  10. 2022Data Feedback Loops: Model-driven Amplification of Dataset Biasescited by 60
  11. 2022Will Large-scale Generative Models Corrupt Future Datasets?cited by 59
  12. 2023Towards Understanding the Interplay of Generative Artificial Intelligence and the Internetcited by 54
  13. 2023Combining Generative Artificial Intelligence (AI) and the Internet: Heading towards Evolution or Degradation?cited by 41
  14. 2024Self-Consuming Generative Models with Curated Data Provably Optimize Human Preferencescited by 39
  15. 2024Self-Correcting Self-Consuming Loops for Generative Model Trainingcited by 37
  16. 2024Towards Theoretical Understandings of Self-Consuming Generative Modelscited by 29
  17. 2023Nepotistically Trained Generative-AI Models Collapsecited by 27
  18. 2024Heat Death of Generative Models in Closed-Loop Learningcited by 22

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 7.

12 papers.

  1. 2025Generating Multi-Image Synthetic Data for Text-to-Image Customizationcited by 25
  2. 2025A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspectivecited by 18
  3. 2025The Intricate Dance of Prompt Complexity, Quality, Diversity, and Consistency in T2I Modelscited by 6
  4. 2026Quantifying Error Propagation and Model Collapse in Diffusion Modelscited by 4
  5. 2025REED-VAE: RE-Encode Decode Training for Iterative Image Editing with Diffusion Modelscited by 3
  6. 2025An Exploration of Default Images in Text-to-Image Generationcited by 1
  7. 2026Banana100: Breaking NR-IQA Metrics by 100 Iterative Image Replications with Nano Banana Procited by 1
  8. 2026Maximum Likelihood Estimation for Entity Ranking under Iterative Synthetic Data Augmentationcited by 1
  9. 2026A Markovian View of Iterative-Feedback Loops in Image Generative Models: Neural Resonance and Model Collapsecited by 0
  10. 2026The evolution of sex for artificial intelligence: a population-genetic framework for multigenerational model populationscited by 0
  11. 2026ReGain: Restoring Subject Fidelity in Personalization on Synthetic Imagescited by 0
  12. 2026Feature Selective Model Collapse in Diffusion Models: Total Replacement versus Fixed-Budget Trainingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Efficient Diffusion Models: A Comprehensive Survey from Principles to Practices
  2. 2023DiffFit: Unlocking Transferability of Large Diffusion Models via Simple Parameter-Efficient Fine-Tuning
  3. 2026Uncovering the Generative Landscape of Diffusion Models: A Spectral-Guided Quality-Diversity Diffusion Framework
  4. 2024Heuristically Adaptive Diffusion-Model Evolutionary Strategy
  5. 2024DiffuseHigh: Training-free Progressive High-Resolution Image Synthesis through Structure Guidance
  6. 2026Diff-ES: Stage-wise Structural Diffusion Pruning via Evolutionary Search
  7. 2026Source-Prior-Driven Selective Adaptation for Efficient Diffusion Model Finetuning
  8. 2024EvolvED: Evolutionary Embeddings to Understand the Generation Process of Diffusion Models