arXiv：cs.CL· Yuxin Meng, Ruixu Zhang, Junjie Wang, Yuhan Suo, Yuhan Sun, Ruining Hu, Yiyao Yu, Yubin Wang, Shouwei Ruan, Bin Wang, Yue Liao, Yuxiang Zhang, Yujiu Yang·· 3 小时前
WebLoop：面向网页生成的执行接地循环学习框架
Learning the Loop, Not Just the Page: Execution-Grounded Loop Learning for Web Generation
AI 导读
针对网页生成中 Critic 环节缺乏可直接执行结果的问题，研究者提出 WebLoop 框架，在共享策略内联合学习生成、批判与修复三个角色。基于 Qwen3.5-9B，WebLoop 在 WebRise 上达到 41.5 Overall，在 WebGen-Bench 上达 38.9% 准确率，分别较基座模型提升 11.3 和 15.4 分。
正文
Authors:Yuxin Meng, Ruixu Zhang, Junjie Wang, Yuhan Suo, Yuhan Sun, Ruining Hu, Yiyao Yu, Yubin Wang, Shouwei Ruan, Bin Wang, Yue Liao, Yuxiang Zhang, Yujiu Yang
Abstract:Functional Web generation is increasingly optimized with executable rewards, yet existing methods largely focus on the quality of the final page and leave the process of diagnosing and repairing imperfect implementations underexplored. We identify a central challenge in this setting: the Generator and Refiner produce executable artifacts with direct environment rewards, whereas the intermediate Critic influences downstream behavior without a directly executable outcome. We introduce WebLoop, an execution-grounded framework that jointly learns generation, critique, and refinement within a shared policy. WebLoop trains an execution-free Critic with complementary signals for requirement-level discriminability and downstream helpfulness, first establishing reliable diagnosis and then introducing consequence-aware credit, while all three roles are jointly optimized with group-relative policy learning. With Qwen3.5-9B, WebLoop reaches 41.5 Overall on WebRise and 38.9% accuracy on WebGen-Bench, improving the base model by 11.3 and 15.4 points, respectively. The gains transfer to first-pass generation, persist at 27B scale, and generalize from text-only training to multimodal inputs. Controlled analyses further show that the improvement cannot be explained by an additional refinement pass alone, highlighting the importance of learning the Critic and the loop itself.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.11543 [cs.CL] |
| (or arXiv:2610.11543v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11543 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yuxin Meng [view email]
[v1]
Thu, 8 Oct 2026 09:11:17 UTC (3,068 KB)
来源：arXiv：cs.CL · arxiv.org