FailureAtlas:Mapping the Failure Landscape of T2I Models via Active Exploration
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914058025828352 |
|---|---|
| author | Chen, Muxi Zhang, Zhaohua Zhao, Chenchen Chen, Mingyang Jiang, Wenyu Jiang, Tianwen Zhuo, Jianhuan Tang, Yu Xiao, Qiuyong Zhang, Jihong Xu, Qiang |
| author_facet | Chen, Muxi Zhang, Zhaohua Zhao, Chenchen Chen, Mingyang Jiang, Wenyu Jiang, Tianwen Zhuo, Jianhuan Tang, Yu Xiao, Qiuyong Zhang, Jihong Xu, Qiang |
| contents | Static benchmarks have provided a valuable foundation for comparing Text-to-Image (T2I) models. However, their passive design offers limited diagnostic power, struggling to uncover the full landscape of systematic failures or isolate their root causes. We argue for a complementary paradigm: active exploration. We introduce FailureAtlas, the first framework designed to autonomously explore and map the vast failure landscape of T2I models at scale. FailureAtlas frames error discovery as a structured search for minimal, failure-inducing concepts. While it is a computationally explosive problem, we make it tractable with novel acceleration techniques. When applied to Stable Diffusion models, our method uncovers hundreds of thousands of previously unknown error slices (over 247,000 in SD1.5 alone) and provides the first large-scale evidence linking these failures to data scarcity in the training set. By providing a principled and scalable engine for deep model auditing, FailureAtlas establishes a new, diagnostic-first methodology to guide the development of more robust generative AI. The code is available at https://github.com/cure-lab/FailureAtlas |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_21995 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | FailureAtlas:Mapping the Failure Landscape of T2I Models via Active Exploration Chen, Muxi Zhang, Zhaohua Zhao, Chenchen Chen, Mingyang Jiang, Wenyu Jiang, Tianwen Zhuo, Jianhuan Tang, Yu Xiao, Qiuyong Zhang, Jihong Xu, Qiang Computer Vision and Pattern Recognition Static benchmarks have provided a valuable foundation for comparing Text-to-Image (T2I) models. However, their passive design offers limited diagnostic power, struggling to uncover the full landscape of systematic failures or isolate their root causes. We argue for a complementary paradigm: active exploration. We introduce FailureAtlas, the first framework designed to autonomously explore and map the vast failure landscape of T2I models at scale. FailureAtlas frames error discovery as a structured search for minimal, failure-inducing concepts. While it is a computationally explosive problem, we make it tractable with novel acceleration techniques. When applied to Stable Diffusion models, our method uncovers hundreds of thousands of previously unknown error slices (over 247,000 in SD1.5 alone) and provides the first large-scale evidence linking these failures to data scarcity in the training set. By providing a principled and scalable engine for deep model auditing, FailureAtlas establishes a new, diagnostic-first methodology to guide the development of more robust generative AI. The code is available at https://github.com/cure-lab/FailureAtlas |
| title | FailureAtlas:Mapping the Failure Landscape of T2I Models via Active Exploration |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.21995 |