# Self-Distillation Policy Optimization via Visual Feedback: Bridging Code and Visual Artifacts

arXiv: https://arxiv.org/abs/2606.10334 · HTML: https://mlsanity.com/paper/2606.10334/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal distillation, one of 12 directions in Knowledge distillation.
- Builds on 31 of its 38 references that are in the map.
- First posted on arXiv 9 Jun 2026.

## Placed in

- Knowledge distillation / [Multimodal distillation](https://mlsanity.com/distill/multimodal-distillation/)

## Builds on (most cited first, 30 of 31)

- [On-Policy Distillation of Language Models: Learning from Self-Generated Mistakes](https://mlsanity.com/paper/2306.13649/) (2023)
- [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://mlsanity.com/paper/2210.03347/) (2022)
- [Reinforcement Learning via Self-Distillation](https://mlsanity.com/paper/2601.20802/) (2026)
- [Unifying distillation and privileged information](https://mlsanity.com/paper/1511.03643/) (2015)
- [A Survey of On-Policy Distillation for Large Language Models](https://mlsanity.com/paper/2604.00626/) (2026)
- [Design2Code: Benchmarking Multimodal Code Generation for Automated Front-End Engineering](https://mlsanity.com/paper/2403.03163/) (2024)
- [Unlocking the conversion of Web Screenshots into HTML Code with the WebSight Dataset](https://mlsanity.com/paper/2403.09029/) (2024)
- [ChartMimic: Evaluating LMM's Cross-Modal Reasoning Capability via Chart-to-Code Generation](https://mlsanity.com/paper/2406.09961/) (2024)
- [pix2code: Generating Code from a Graphical User Interface Screenshot](https://mlsanity.com/paper/1705.07962/) (2017)
- [ChartCoder: Advancing Multimodal Large Language Model for Chart-to-Code Generation](https://mlsanity.com/paper/2501.06598/) (2025)
- [Privileged Information Distillation for Language Models](https://mlsanity.com/paper/2602.04942/) (2026)
- [PPTAgent: Generating and Evaluating Presentations Beyond Text-to-Slides](https://mlsanity.com/paper/2501.03936/) (2025)
- [Web2Code: A Large-scale Webpage-to-Code Dataset and Evaluation Framework for Multimodal LLMs](https://mlsanity.com/paper/2406.20098/) (2024)
- [Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers](https://mlsanity.com/paper/2505.21497/) (2025)
- [WebCode2M: A Real-World Dataset for Code Generation from Webpage Designs](https://mlsanity.com/paper/2404.06369/) (2024)
- [HDPO: Hybrid Distillation Policy Optimization via Privileged Self-Distillation](https://mlsanity.com/paper/2603.23871/) (2026)
- [Torch.fx: Practical Program Capture and Transformation for Deep Learning in Python](https://mlsanity.com/paper/2112.08429/) (2021)
- [Rendering-Aware Reinforcement Learning for Vector Graphics Generation](https://mlsanity.com/paper/2505.20793/) (2025)
- [GATES: Self-Distillation under Privileged Context with Consensus Gating](https://mlsanity.com/paper/2602.20574/) (2026)
- [Learning Only with Images: Visual Reinforcement Learning with Reasoning, Rendering, and Visual Feedback](https://mlsanity.com/paper/2507.20766/) (2025)
- [ScreenCoder: Advancing Visual-to-Code Generation for Front-End Automation via Modular Multimodal Agents](https://mlsanity.com/paper/2507.22827/) (2025)
- [Respecting Self-Uncertainty in On-Policy Self-Distillation for Efficient LLM Reasoning](https://mlsanity.com/paper/2605.13255/) (2026)
- [VinciCoder: Unifying Multimodal Code Generation via Coarse-to-fine Visual Reinforcement Learning](https://mlsanity.com/paper/2511.00391/) (2025)
- [Breaking the SFT Plateau: Multimodal Structured Reinforcement Learning for Chart-to-Code Generation](https://mlsanity.com/paper/2508.13587/) (2025)
- [ChartMaster: Advancing Chart-to-Code Generation with Real-World Charts and Chart Similarity Reinforcement Learning](https://mlsanity.com/paper/2508.17608/) (2025)
- [UI2Code^N: UI-to-Code Generation as Interactive Visual Optimization](https://mlsanity.com/paper/2511.08195/) (2025)
- [ReLook: Vision-Grounded RL with a Multimodal LLM Critic for Agentic Web Coding](https://mlsanity.com/paper/2510.11498/) (2025)
- [Anti-Self-Distillation for Reasoning RL via Pointwise Mutual Information](https://mlsanity.com/paper/2605.11609/) (2026)
- [DesignCoder: Hierarchy-Aware and Self-Correcting UI Code Generation with Large Language Models](https://mlsanity.com/paper/2506.13663/) (2025)
- [ChartEditor: A Reinforcement Learning Framework for Robust Chart Editing](https://mlsanity.com/paper/2511.15266/) (2025)

## Built on by (most cited first, 1 of 1)

- [MIRROR: From Imitation to Internalization in LLM Personalization](https://mlsanity.com/paper/2610.09795/) (2026)

## Abstract (arXiv)

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.