GDB: GraphicDesignBench

39 benchmarks for evaluating vision-language models on graphic design tasks — layout, typography, SVG, template matching, animation. Built on 1,148 real design layouts from the Lica dataset.

Paper: arXiv:2604.04192  |  Code: github.com/lica-world/GDB  |  Blog: lica.world

Usage

from datasets import load_dataset

ds = load_dataset("lica-world/GDB", "svg-1")

Schema

Field	Type	Description
sample_id	string	Sample identifier
benchmark_id	string	e.g. svg-1, typography-3
domain	string	layout, typography, svg, template, temporal, category, lottie
task_type	string	understanding or generation
prompt	string	Evaluation prompt
ground_truth	string	Expected answer (JSON for complex types)
image	Image	Input image (when applicable)
metadata	string	Task-specific fields as JSON

Evaluation

pip install git+https://github.com/lica-world/GDB.git

from gdb.registry import BenchmarkRegistry

registry = BenchmarkRegistry()
registry.discover()
bench = registry.get("svg-1")
scores = bench.evaluate(predictions, ground_truth)

Citation

@article{gdb2026,
  title={GDB: A Real-World Benchmark for Graphic Design},
  author={Deganutti, Adrienne and Hirsch, Elad and Zhu, Haonan and Seol, Jaejung and Mehta, Purvanshi},
  journal={arXiv preprint arXiv:2604.04192},
  year={2026}
}

Apache 2.0