Scientific Idea Generation (n=96)
0.99GroundingEvoIdeator
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| EvoIdeatorStep=Direct Generation Step2026.03 | 0.99 | 0.19 | 0.06 | 0.03 | 0.18 | 0.96 | 0.32 | 0.37 | |
| EvoIdeatorStep=Feedback Refinement Step2026.03 | 0.99 | 0.31 | 0.94 | 0.35 | 0.58 | 0.99 | 0.47 | 0.18 | |
| Qwen-4BStep=Feedback Refinement Step2026.03 | 0.94 | 0.25 | 0.91 | 0.19 | 0.39 | 0.92 | 0.41 | 0.33 | |
| DS v3.2Step=Feedback Refinement Step2026.03 | 0.91 | 0.54 | 0.9 | 0.3 | 0.72 | 0.88 | 0.42 | 0.25 | |
| Gemini 3 FlashStep=Feedback Refinement Step2026.03 | 0.91 | 0.53 | 0.9 | 0.16 | 0.48 | 0.97 | 0.6 | 0.5 | |
| Gemini 3 FlashStep=Direct Generation Step2026.03 | 0.87 | 0.2 | 0.07 | 0 | 0.1 | 0.9 | 0.55 | 0.81 | |
| Qwen-4BStep=Direct Generation Step2026.03 | 0.85 | 0.09 | 0.05 | 0.01 | 0.08 | 0.75 | 0.21 | 0.41 | |
| DS v3.2Step=Direct Generation Step2026.03 | 0.8 | 0.19 | 0.1 | 0 | 0.12 | 0.72 | 0.31 | 0.4 | |
| DS R1 DistillStep=Feedback Refinement Step2026.03 | 0.59 | 0.09 | 0.63 | 0.06 | 0.34 | 0.59 | 0.16 | 0.03 | |
| DS R1 DistillStep=Direct Generation Step2026.03 | 0.53 | 0 | 0 | 0 | 0 | 0.19 | 0 | 0.16 |