DataEvolver: Let Your Data Build and Improve Itself via Goal-Driven Loop Agents
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917456270852096 |
|---|---|
| author | Zhang, Qisong Wu, Wenzhuo Jia, Zhuangzhuang Yang, Yunhao Zhang, Huayu Zang, Xianghao He, Zhixiang He, Zhongjiang Liang, Kongming Ma, Zhanyu |
| author_facet | Zhang, Qisong Wu, Wenzhuo Jia, Zhuangzhuang Yang, Yunhao Zhang, Huayu Zang, Xianghao He, Zhixiang He, Zhongjiang Liang, Kongming Ma, Zhanyu |
| contents | Constructing controllable visual data is a major bottleneck for image editing and multimodal understanding. Useful supervision is rarely produced by a single rendering pass; instead it emerges through iterative generation, inspection, correction, filtering, and export. We present DataEvolver, a closed-loop visual data engine that organizes this process around explicit goals, persistent artifacts, bounded corrective actions, and acceptance decisions. DataEvolver supports multiple artifact types, including RGB images, masks, depth maps, normal maps, meshes, poses, trajectories, and review traces. In the current release, the system operates through two coupled loops: generation-time self-correction within each sample and validation-time self-expansion across dataset rounds. We validate the framework on an image-level object-rotation setting. With a fixed Qwen-Edit LoRA probe, our final Ours+DualGate model outperforms both the unadapted base model and a public multi-angle LoRA on SpatialEdit and a held-out evaluation set. Ablations show a consistent improvement path from scene-aware generation to feedback-driven correction and dual-gated validation. Beyond the released rotation data, our main contribution is a reusable framework for building visual datasets through explicit goal tracking, review, correction, and acceptance loops. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_01789 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | DataEvolver: Let Your Data Build and Improve Itself via Goal-Driven Loop Agents Zhang, Qisong Wu, Wenzhuo Jia, Zhuangzhuang Yang, Yunhao Zhang, Huayu Zang, Xianghao He, Zhixiang He, Zhongjiang Liang, Kongming Ma, Zhanyu Artificial Intelligence Constructing controllable visual data is a major bottleneck for image editing and multimodal understanding. Useful supervision is rarely produced by a single rendering pass; instead it emerges through iterative generation, inspection, correction, filtering, and export. We present DataEvolver, a closed-loop visual data engine that organizes this process around explicit goals, persistent artifacts, bounded corrective actions, and acceptance decisions. DataEvolver supports multiple artifact types, including RGB images, masks, depth maps, normal maps, meshes, poses, trajectories, and review traces. In the current release, the system operates through two coupled loops: generation-time self-correction within each sample and validation-time self-expansion across dataset rounds. We validate the framework on an image-level object-rotation setting. With a fixed Qwen-Edit LoRA probe, our final Ours+DualGate model outperforms both the unadapted base model and a public multi-angle LoRA on SpatialEdit and a held-out evaluation set. Ablations show a consistent improvement path from scene-aware generation to feedback-driven correction and dual-gated validation. Beyond the released rotation data, our main contribution is a reusable framework for building visual datasets through explicit goal tracking, review, correction, and acceptance loops. |
| title | DataEvolver: Let Your Data Build and Improve Itself via Goal-Driven Loop Agents |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2605.01789 |