arXiv cs.AI· Jiawen Li, Chenming Ge, Alice Gao, Rishab Vempati, Yi-Hao Peng, Jason Wu·· 12 小时前AI 评分31
RELIEF：利用交互体验与反馈的修订式学习提升代码生成模型
Revision-Based Learning from Interactive Experience and Feedback
AI 导读
RELIEF 是一种利用生成界面的反馈改进代码生成模型的方法：先用编码智能体修订界面，再把改进并入原始生成轨迹以构造微调监督。研究收集了 200 个合成网站、355 段交互录制和 24 位用户的反馈，用代码检查、自动交互、显式人类反馈与人类交互录制四类批评微调两个代码生成模型的变体。结果显示反馈效果随基座模型能力而异，其中从人类交互录制推断的批评在测试模型下带来最大的可用性提升。
正文 · 原文
Abstract:Code generation models are increasingly used to produce interactive artifacts, such as websites and user interfaces. Many properties that determine whether these interfaces are useful only become apparent through realistic interaction, making them difficult to incorporate into model training. In this paper, we introduce RELIEF (REvision-based Learning from Interactive Experience and Feedback), a method for improving code generation models using feedback on their generated interfaces. RELIEF uses a coding agent to revise these interfaces, then incorporates the improvements into their original generation traces to produce supervision for fine-tuning. To evaluate RELIEF, we collected a dataset of 200 synthetically generated websites paired with 355 interaction recordings and accompanying feedback from 24 users. We used critiques derived from code inspection, automated interaction, explicit human feedback, and human interaction recordings to fine-tune variants of two code generation models. Our results show that feedback effectiveness varies with base model capabilities, with critiques inferred from human interaction recordings leading to the largest usability improvements overall under our tested models.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.26636 [cs.AI] |
| (or arXiv:2610.26636v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.26636 arXiv-issued DOI via DataCite |
Submission history
From: Jiawen Li [view email]
[v1]
Mon, 19 Oct 2026 16:01:17 UTC (6,848 KB)
[v2]
Tue, 20 Oct 2026 16:11:17 UTC (6,775 KB)
来源：arXiv cs.AI · arxiv.org