RULER: Instance-aware Rubric Rewards for SVG Generation

Researchers introduced RULER (Instance-aware Rubric Rewards for Reinforcement Learning), a method that converts each natural-language instruction into an instance-aware rubric of six items spanning semantic, visual, and stylistic axes, with a judge VLM scoring rendered rollouts item-by-item and weighted satisfactions forming a reward optimized via Group Relative Policy Optimization. On MMSVG-Illustration and MMSVG-Icon, RULER lifted the rubric score from 0.432/0.395 to 0.693/0.683, surpassing dedicated SVG specialists and matching the substantially larger DeepSeek-V3, without requiring paired SVG ground truth or human preference labels. The work is published on arXiv as 2609.25270v1.

arXiv:2609.25270v1 Announce Type: new Abstract: Generating Scalable Vector Graphics SVG code from natural-language instructions is an open-ended task without absolute visual ground truth, leaving both evaluation and policy optimization without a faithful signal. Scalar metrics CLIP, Aesthetic calibrated on natural images transfer poorly to stylized vector content, and reusing them as RL rewards triggers reward hacking. We address both limitations with rubric-based scoring. We first establish empirically that prompting a vision-language judge with a multi-axis rubric correlates with human judgments far better than scalar metrics, both across samples and within instructions. Building on this finding, we introduce RULER Instance-aware Rubric Rewards for Reinforcement Learning , which converts each instruction into an instance-aware rubric of six items spanning semantic, visual, and stylistic axes; a judge VLM scores rendered rollouts item-by-item, and the weighted satisfactions form a fine-grained reward optimized via Group Relative Policy Optimization. Because the rubric is derived from text alone, RULER requires neither paired SVG ground truth nor human preference labels. On MMSVG-Illustration and MMSVG-Icon, RULER lifts the rubric score from 0.432/0.395 to 0.693/0.683, surpassing dedicated SVG specialists and matching the substantially larger DeepSeek-V3, with ablations identifying rubric design as the active lever for RL on open-ended SVG generation. The project page is available at https://hangyuran.github.io/RULER/.