Scientific data visualization on MatPlotBench
56.73ScoreMatPlotAgent
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MatPlotAgentBackbone=GPT-4, Visual Agent=Gemini Pro Vision2024.02 | 56.73 | 7.87 | |
| GPT-4Evaluation Protocol=Direct Decoding2024.02 | 48.86 | — | |
| MatPlotAgentBackbone=GPT-3.5, Visual Agent=Gemini Pro Vision2024.02 | 43.48 | 5.45 | |
| GPT-3.5Evaluation Protocol=Direct Decoding2024.02 | 38.03 | — |