Deciphering Personalization: Towards Fine-Grained Explainability in Natural Language for Personalized Image Generation Models
Not on the map. Why?
Abstract (arXiv)
Image generation models are usually personalized in practical uses in order to better meet the individual users' heterogeneous needs, but most personalized models lack explainability about how they are being personalized. Such explainability can be provided via visual features in generated images, but is difficult for human users to understand. Explainability in natural language is a better choice, but the existing approaches to explainability in natural language are limited to be coarse-grained. They are unable to precisely identify the multiple aspects of personalization, as well as the varying levels of personalization in each aspect. To address such limitation, in this paper we present a new technique, namely FineXL, towards Fine-grained eXplainability in natural Language for personalized image generation models. FineXL can provide natural language descriptions about each distinct aspect of personalization, along with quantitative scores indicating the level of each aspect of personalization. Experiment results show that FineXL can improve the accuracy of explainability by 56%, when different personalization scenarios are applied to multiple types of image generation models.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 56 of its 69 references that are in the map. First posted on arXiv 2 Nov 2025.
Builds on
56 of 69 references.
References · 56
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
- 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
- 2023GPT-4 Technical Reportcited by 18,396
- 2014Microsoft COCO: Common Objects in Contextcited by 17,608
- 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
- 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,585
- 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
- 2022Classifier-Free Diffusion Guidancecited by 5,847
- 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
- 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,181
- 2021Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervisioncited by 3,864
- 2022DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generationcited by 3,300
38 more
- 2022An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversioncited by 2,245
- 2017Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)cited by 1,319
- 2022Reproducible scaling laws for contrastive language-image learningcited by 1,173
- 2024Chameleon: Mixed-Modal Early-Fusion Foundation Modelscited by 927
- 2023EVA-CLIP: Improved Training Techniques for CLIP at Scalecited by 798
- 2023The Linear Representation Hypothesis and the Geometry of Large Language Modelscited by 716
- 2018What you can cram into a single vector: Probing sentence embeddings for linguistic propertiescited by 457
- 2022Dataset Distillation by Matching Training Trajectoriescited by 442
- 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 422
- 2021Dataset Condensation with Distribution Matchingcited by 365
- 2020Parameter-Efficient Transfer Learning with Diff Pruningcited by 352
- 2020Explaining by Removing: A Unified Framework for Model Explanationcited by 218
- 2023Style Aligned Image Generation via Shared Attentioncited by 216
- 2019Controllable Text-to-Image Generationcited by 187
- 2023Linear Representations of Sentiment in Large Language Modelscited by 163
- 2018StoryGAN: A Sequential Conditional GAN for Story Visualizationcited by 138
- 2023Taming the Power of Diffusion Models for High-Quality Virtual Try-On with Appearance Flowcited by 124
- 2024ANOLE: An Open, Autoregressive, Native Large Multimodal Models for Interleaved Image-Text Generationcited by 114
- 2022VectorFusion: Text-to-SVG by Abstracting Pixel-Based Diffusion Modelscited by 112
- 2024Can LLM be a Personalized Judge?cited by 100
- 2024IMAGDressing-v1: Customizable Virtual Dressingcited by 100
- 2025SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoderscited by 80
- 2024On the Origins of Linear Representations in Large Language Modelscited by 79
- 2022Impossibility Theorems for Feature Attributioncited by 78
- 2023Goal Driven Discovery of Distributional Differences via Language Descriptionscited by 62
- 2017Towards Diverse and Natural Image Descriptions via a Conditional GANcited by 59
- 2022Describing Differences between Text Distributions with Natural Languagecited by 54
- 2023Describing Differences in Image Sets with Natural Languagecited by 53
- 2023Linear Spaces of Meanings: Compositional Structures in Vision-Language Modelscited by 48
- 2023RenAIssance: A Survey into AI Text-to-Image Generation in the Era of Large Modelcited by 47
- 2023NoiseCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions in Diffusion Modelscited by 47
- 2024Decomposing The Dark Matter of Sparse Autoencoderscited by 45
- 2023Bridging the Human-AI Knowledge Gap: Concept Discovery and Transfer in AlphaZerocited by 42
- 2023Changes to Captions: An Attentive Network for Remote Sensing Change Captioningcited by 41
- 2022Image Difference Captioning with Pre-training and Contrastive Learningcited by 39
- 2024Learning Interpretable Concepts: Unifying Causal Representation Learning and Foundation Modelscited by 29
- 2023Rejuvenating image-GPT as Strong Visual Representation Learnerscited by 16
- 2022GSCLIP : A Framework for Explaining Distribution Shifts in Natural Languagecited by 14
50 most cited of 56 All 56 (JSON)
Built on by
Citing papers per year: 2026: 1.
1 paper.