mlsanityData to 14 Oct 2026

T2I-ConBench: Text-to-Image Benchmark for Continual Post-training

arXiv 2505.16875 ↗First posted 22 May 2025cs.CV cs.AI cs.LGcited by 1

On the map: Diffusion models → Image generation, text-to-image and editing

Abstract (arXiv)

Continual post-training adapts a single text-to-image diffusion model to learn new tasks without incurring the cost of separate models, but naive post-training causes forgetting of pretrained knowledge and undermines zero-shot compositionality. We observe that the absence of a standardized evaluation protocol hampers related research for continual post-training. To address this, we introduce T2I-ConBench, a unified benchmark for continual post-training of text-to-image models. T2I-ConBench focuses on two practical scenarios, item customization and domain enhancement, and analyzes four dimensions: (1) retention of generality, (2) target-task performance, (3) catastrophic forgetting, and (4) cross-task generalization. It combines automated metrics, human-preference modeling, and vision-language QA for comprehensive assessment. We benchmark ten representative methods across three realistic task sequences and find that no approach excels on all fronts. Even joint "oracle" training does not succeed for every task, and cross-task generalization remains unsolved. We release all datasets, code, and evaluation tools to accelerate research in continual post-training for text-to-image models.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models. Builds on 65 of its 79 references that are in the map. First posted on arXiv 22 May 2025.

Builds on

65 of 79 references.

Closest ancestors · 8

Same direction, newest first

  1. 2025BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset14 May
  2. 2024Progressive Compositionality in Text-to-Image Generative Models22 Oct
  3. 2024A Survey on Personalized Content Synthesis with Diffusion Models9 May
  4. 2023GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment17 Oct
  5. 2023Create Your World: Lifelong Text-to-Image Diffusion8 Sep
  6. 2023Continual Diffusion: Continual Customization of Text-to-Image Diffusion with C-LoRA12 Apr
  7. 2022DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation25 Aug
  8. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding23 May

Other references · 57

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  5. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  6. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  7. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
  8. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  9. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  10. 2022Scalable Diffusion Models with Transformerscited by 6,054
  11. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
  12. 2024Qwen2.5 Technical Reportcited by 4,942
38 more
  1. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  2. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,356
  3. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  4. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,236
  5. 2023MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Modelscited by 2,711
  6. 2022Prompt-to-Prompt Image Editing with Cross Attention Controlcited by 2,431
  7. 2019ZeRO: Memory Optimizations Toward Training Trillion Parameter Modelscited by 1,301
  8. 2023ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generationcited by 1,045
  9. 2021Towards a Unified View of Parameter-Efficient Transfer Learningcited by 872
  10. 2023A Comprehensive Survey of Continual Learning: Theory, Method and Applicationcited by 872
  11. 2023PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesiscited by 864
  12. 2023The Dawn of LMMs: Preliminary Explorations with GPT-4V(ision)cited by 681
  13. 2024PixArt-Sigma: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generationcited by 329
  14. 2023Orthogonal Subspace Learning for Language Model Continual Learningcited by 312
  15. 2023Human Preference Score: Better Aligning Text-to-Image Models with Human Preferencecited by 287
  16. 2023Emu: Enhancing Image Generation Models Using Photogenic Needles in a Haystackcited by 279
  17. 2024Continual Learning of Large Language Models: A Comprehensive Surveycited by 234
  18. 2018A Note on the Inception Scorecited by 226
  19. 2023Continual Pre-training of Language Modelscited by 198
  20. 2024DreamBench++: A Human-Aligned Benchmark for Personalized Image Generationcited by 131
  21. 2024Learning Dynamics of LLM Finetuningcited by 95
  22. 2023Compositional Abilities Emerge Multiplicatively: Exploring Diffusion Models on a Synthetic Taskcited by 93
  23. 2020Improving Image Captioning with Better Use of Captionscited by 84
  24. 2024Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilitiescited by 69
  25. 2023TRACE: A Comprehensive Benchmark for Continual Learning in Large Language Modelscited by 66
  26. 2020Conditionally Adaptive Multi-Task Learning: Improving Transfer Learning in NLP Using Fewer Parameters & Less Datacited by 50
  27. 2022Continual Training of Language Models for Few-Shot Learningcited by 33
  28. 2024A Statistical Theory of Regularization-Based Continual Learningcited by 33
  29. 2023DiffPortrait3D: Controllable Diffusion for Zero-Shot Portrait View Synthesiscited by 26
  30. 2023Robust Visual Question Answering: Datasets, Methods, and Future Challengescited by 23
  31. 2023DiffBoost: Enhancing Medical Image Segmentation via Text-Guided Diffusion Modelcited by 23
  32. 2024HFT: Half Fine-Tuning for Large Language Modelscited by 18
  33. 2022A Theory for Knowledge Transfer in Continual Learningcited by 16
  34. 2023DomainStudio: Fine-Tuning Diffusion Models for Domain-Driven Image Generation using Limited Datacited by 15
  35. 2024MoFO: Momentum-Filtered Optimizer for Mitigating Forgetting in LLM Fine-Tuningcited by 14
  36. 2024Customizing Text-to-Image Diffusion with Object Viewpoint Controlcited by 13
  37. 2018Approximate Fisher Information Matrix to Characterise the Training of Deep Neural Networkscited by 12
  38. 2023Diffusion Based Augmentation for Captioning and Retrieval in Cultural Heritagecited by 12

50 most cited of 57 All 57 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026ACE-LoRA: Adaptive Orthogonal Decoupling for Continual Image Editingcited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2024Caption, Create, Continue: Continual Learning with Pre-trained Generative Vision-Language Models
  2. 2022The CLEAR Benchmark: Continual LEArning on Real-World Imagery
  3. 2025Synthetic Data is an Elegant GIFT for Continual Vision-Language Models
  4. 2025Continual Unlearning for Text-to-Image Diffusion Models: A Regularization Perspective
  5. 2025Harnessing Textual Semantic Priors for Knowledge Transfer and Refinement in CLIP-Driven Continual Learning
  6. 2022Human Evaluation of Text-to-Image Models on a Multi-Task Benchmark
  7. 2023InstructCV: Instruction-Tuned Text-to-Image Diffusion Models as Vision Generalists
  8. 2023CTP: Towards Vision-Language Continual Pretraining via Compatible Momentum Contrast and Topology Preservation