通过视觉反馈的自蒸馏策略优化：连接代码与视觉工件
Self-Distillation Policy Optimization via Visual Feedback: Bridging Code and Visual Artifacts
AI总结：
提出Visual-SDPO框架，利用渲染视觉反馈作为特权上下文，通过自蒸馏和视觉引导的代码信用加权优化代码生成视觉工件的质量，在图表、UI和幻灯片生成任务上显著提升性能。
摘要：
代码生成大语言模型（LLMs）通过编写由不可微渲染器执行的程序，越来越多地生成图表、网页和幻灯片等视觉工件，在观察渲染结果之前就确定了代码。因此，原本可执行的代码常常产生具有视觉显著缺陷的工件，包括元素重叠、文本裁剪、对齐破坏、对比度低和溢出。我们研究针对代码生成视觉工件的视觉反馈自蒸馏。我们提出Visual-SDPO，一种自蒸馏策略优化框架，将渲染的视觉反馈视为权重共享教师的特权上下文，并将该反馈蒸馏到编码学生中。为了使监督具有空间针对性而非均匀性，我们引入视觉引导的代码信用加权，将每个检测到的缺陷追溯到影响该元素的代码语句，并放大这些语句上的蒸馏信号。序列级GRPO（组相对策略优化）项通过奖励可执行、视觉质量高的 rollout 来补充密集的 token 级目标，而失败的执行通过自蒸馏路径仍然可学习，通过将执行错误作为特权上下文传递给教师。我们使用统一的 Qwen3-VL-8B-Instruct 骨干网络，在图表、网页/UI和幻灯片生成任务上实例化 Visual-SDPO。在图表到代码、UI到代码和幻灯片生成基准（ChartMimic、Design2Code和AeSlides）上，Visual-SDPO 在主要指标上比零样本基线提升超过10个绝对点，比GRPO提升至少2.4个点，且训练步骤更少，无额外推理成本。
英文摘要：
Code-generating large language models (LLMs) increasingly produce visual artifacts such as charts, web pages, and slides by writing programs that are executed by non-differentiable renderers, committing to code before observing the render. As a result, otherwise executable code often yields artifacts with visually salient defects, including overlapping elements, clipped text, broken alignment, low contrast, and overflow. We study visual-feedback self-distillation for code-generated visual artifacts. We propose Visual-SDPO, a self-distillation policy-optimization framework that treats rendered visual feedback as privileged context for a weight-sharing teacher and distills this feedback into a coding student. To make supervision spatially targeted rather than uniform, we introduce Visual-Grounded Code Credit Weighting, which traces each detected defect back to the code statements responsible for the affected elements and amplifies the distillation signal on those statements. A sequence-level GRPO (Group Relative Policy Optimization) term complements the dense token-level objective by rewarding executable, visually high-quality rollouts, while failed executions remain learnable through the self-distillation path by passing execution errors as privileged context to the teacher. We instantiate Visual-SDPO for chart, web/UI, and slide generation with a unified Qwen3-VL-8B-Instruct backbone. Across chart-to-code, UI-to-code, and slide-generation benchmarks (ChartMimic, Design2Code, and AeSlides), Visual-SDPO improves over the zero-shot base by more than 10 absolute points in the primary metric and over GRPO by at least 2.4 points, with fewer training steps and no added inference-time cost.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
