Reliable Reasoning in SVG-LLMs via Multi-Task Multi-Reward Reinforcement Learning

Haomin Wang; Qi Wei; Qianli Ma; Shengyuan Ding; Jinhui Yin; Kai Chen; Hongjie Zhang

cs.CVMar 17, 2026

Reliable Reasoning in SVG-LLMs via Multi-Task Multi-Reward Reinforcement Learning

Authors: Haomin Wang, Qi Wei, Qianli Ma, Shengyuan Ding, Jinhui Yin, Kai Chen, Hongjie Zhang

Organizations: Shanghai Jiao Tong University · Shanghai AI Laboratory · Nanjing University · Fudan University

Abstract

With the rapid advancement of vision-language models, an increasing number of studies have explored their potential for SVG generation tasks. Although existing approaches improve performance by constructing large-scale SVG datasets and introducing SVG-specific tokens, they still suffer from limited generalization, redundant paths in code outputs, and a lack of explicit reasoning. In this work, we present CTRL-S (Chain-of-Thought Reinforcement Learning for SVG), a unified framework that introduces a chain-of-thought mechanism to explicitly expose the model's reasoning process during SVG generation. To support this structured reasoning, we construct SVG-Sophia, a high-quality dataset containing 145K samples across SVG code refinement, Text-to-SVG, and Image-to-SVG tasks. By training the model to generate group-level structured SVG code, CTRL-S significantly improves structural coherence and visual fidelity. Furthermore, we adopt the GRPO algorithm and design a multi-reward optimization framework, incorporating DINO, image-text similarity, format, and code efficiency rewards. Through joint multi-reward optimization and multi-task training, our approach systematically enhances overall generation capabilities. Extensive experiments show that CTRL-S outperforms existing methods, achieving higher task success rates, superior SVG code quality, and exceptional visual fidelity.

Figures & tables

Figure 1: Overview of CTRL-S. (Top Left) The Multi-Task Multi-Reward GRPO training framework integrates diverse generation tasks (Text-to-SVG, Image-to-SVG, and Code Refinement) guided by multiple rewards. (Bottom Left) During inference, CTRL-S leverages chain-of-thought reasoning to plan step-by-step drawing operations before generating the final group-level structured SVG code, ensuring a clear one-to-one correspondence between the reasoning steps and the generated code groups. (Right) Examples of high-quality generated SVGs and successful code refinement processes.

Figure 2: The overall pipeline of CTRL-S. (1) Two-Stage SFT: The model is first trained on 1M SAgoge samples to align SVG-specific tokens, and then fine-tuned on SVG-Sophia to learn CoT-structured responses with explicit step-wise planning. (2) Multi-Task Multi-Reward RL: We jointly optimize Text-to-SVG, Image-to-SVG, and SVG refinement tasks via a multi-reward mechanism, including Format Reward, DINO Reward, Image-text Similarity Reward, and Code Efficiency Reward, to improve structural validity, visual fidelity, semantic alignment, and concise code generation.

Traditional SVG Methods
Model	Data	Text-to-SVG					Image-to-SVG
Model	Data	FID $\downarrow$	CLIP-T2I $\uparrow$	CLIP-I2I $\uparrow$	SR	Tokens	DINO $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$	SR	Tokens
IconShop	–	32.288	20.894	70.922	86.51%	1.6k	–	–	–	–	–
VectorFusion	–	16.594	22.992	70.308	100.0%	33k	–	–	–	–	–
SVGDreamer	–	26.612	20.329	69.975	100.0%	132k	–	–	–	–	–
DiffVG	–	–	–	–	–	–	0.869	0.927	0.097	100.0%	17k

Table 1: SVG generation results on SArena-Icon. SR denotes Success Rate. CTRL-S (SFT) denotes the model after two-stage SFT, CTRL-S (SFT+RL) denotes the full model after RL post-training.

Model	DINO $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$	SR	Tokens
Vision-Language Models
Llama 4 Scout	0.840	0.634	0.383	97.00%	788
Llama 4 Maverick	0.845	0.615	0.377	94.75%	616
Qwen3-VL-235B-A22B	0.809	0.515	0.403	77.84%	840
GPT-5.2	0.911	0.640	0.342	99.26%	975
Claude-Sonnet-4.5	0.796	0.579	0.401	84.58%	790

Table 2: SVG refinement results on SVG-Sophia Code Refinement Benchmark.

Panel of Figure 3 — Figure 3: Qualitative comparisons of SVG generation and code refinement between baselines and CTRL-S.

Figure 4: Qualitative visualization of SVG generation quality across RL training steps.

Table 3: Ablation studies on CoT, reward signals, reward ratios, and multi-task training.

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Category	Tokens
Root	<svg, </svg>, <defs, </defs>, <use, </use> ,/>
Grouping	<g, </g>
Shapes	<path, </path>, <rect, </rect>, <circle, </circle>, <ellipse, </ellipse>, <line, </line>, <polyline, </polyline>, <polygon, </polygon>
Text	<text, </text>, <tspan, </tspan>, <textPath, </textPath>
Gradients	<linearGradient, </linearGradient>, <radialGradient, </radialGradient>, <stop, </stop>
Clipping	<clipPath, </clipPath>, <mask, </mask>

Appendix

Table 4: SVG-specific tokens used in CTRL-S.

Figure 5: Examples of Text-to-SVG in SVG-Sophia.

Figure 6: Examples of Image-to-SVG in SVG-Sophia.

Figure 7: Examples of SVG code refinement in SVG-Sophia.

Text-to-SVG Generation

See all 13 papers

RefineSVG: Visual Feedback-Driven Reinforcement Learning for Image-to-SVG Generation
Jul 30, 2026Shaobo Liu, Feiqiao Mao, Shuaishuai Zhou +4Image Generation Large Vision-Language Models
Shenzhen University · Shenzhen, China · Peking University +1
IntroSVG: Learning from Rendering Feedback for Text-to-SVG Generation via an Introspective Generator-Critic Framework
Mar 10, 2026Feiyu Wang, Jiayuan Yang, Zhiyuan Zhao +4Text-to-SVG Generation Vision-Language Models
Fudan University · TeleAI · Northwestern Polytechnical University +1
Evaluating Constrained Iterative Refinement for Scalable Vector Graphics Generation with Off-the-Shelf VLMs
Aug 26, 2026Matthew Perlman, James Beetham, Niels Da Vitoria Lobo +2Vector Graphics Generation VLM Adaptation
University of Massachusetts Amherst · University of Central Florida

Jul 30, 2026J/K move · Enter open · S save

RefineSVG: Visual Feedback-Driven Reinforcement Learning for Image-to-SVG Generation

Shaobo Liu, Feiqiao Mao, Shuaishuai Zhou, Yan Zhan+3

Shenzhen University · Shenzhen, China · Peking University+1

Image Generation Large Vision-Language Models RL for VLMs

Open paper AskPDF arXiv

We propose RefineSVG, a single-step closed-loop visual feedback framework that enables multimodal large language models (MLLMs) to perform high-fidelity image-to-SVG generation through self-correction. Existing MLLM-based approaches rely on single-pass open-loop inference, where the model receives visual input only once and must generate thousands of SVG code tokens without intermediate verification. This paradigm inevitably leads to geometric drift, error accumulation, and visual hallucination on complex images. RefineSVG overcomes this limitation by invoking an external rendering engine after an initial SVG generation pass to compare the rendered output against the target image. The comparison yields a multi-dimensional visual residual map (Diff-Map) that is fed back to the model as a ReAct-style correction signal, driving a targeted correction step. To support this render-observe-correct interaction, we further introduce an SVG-oriented semantic vocabulary that compresses token sequences by over 52%. A progressive training pipeline spanning supervised fine-tuning, rejection-sampling cold-start data construction, and end-to-end agentic reinforcement learning aligns the model with closed-loop visual correction. Extensive experiments show that RefineSVG consistently outperforms existing baselines in reconstruction fidelity, structural accuracy, and code efficiency.Code is available at https://github.com/liuxiaobo66/RefineSVG.

Abstract

Text-to-SVG Generation

Explore similar work

RefineSVG: Visual Feedback-Driven Reinforcement Learning for Image-to-SVG Generation

IntroSVG: Learning from Rendering Feedback for Text-to-SVG Generation via an Introspective Generator-Critic Framework

Evaluating Constrained Iterative Refinement for Scalable Vector Graphics Generation with Off-the-Shelf VLMs