Correlation with Human Judgment on NLV and ChartLLM (171 sampled examples)
0.73Pearson CorrelationMatPlotBench
Evaluation Results
| Method | Links | |
|---|---|---|
| MatPlotBenchScope=All examples2026.01 | 0.73 | |
| Vision ScoreScope=All examples2026.01 | 0.71 | |
| Spec ScoreScope=Vega-Lite examples only2026.01 | 0.65 | |
| SEVQScope=All examples2026.01 | 0.46 |