Symbolic Regression on NewtonBench Overall
67.6Symbolic AccuracyLLM-AutoSciLab
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LLM-AutoSciLabED (Experiment Design)=true2026.05 | 67.6 | 81.5 | 0.15 | |
| Bayesian OptimizationED (Experiment Design)=true2026.05 | 24.54 | 68.52 | 0.334 | |
| PySRED (Experiment Design)=false2026.05 | 24.07 | 74.54 | 0.182 | |
| Bayesian Experimental DesignED (Experiment Design)=true2026.05 | 11.11 | 63.86 | 0.577 | |
| Code-assisted LLMED (Experiment Design)=true2026.05 | 7.41 | 10.19 | 4.912 | |
| LLM-onlyED (Experiment Design)=true2026.05 | 6.48 | 7.41 | 5.039 |