Scientific Idea Generation on AI-Scientist
4.1Absolute NoveltyOur Method - BoN(10)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Our Method - BoN(10)Backbone=Qwen 2.5 14B, Quantization=4-bit, Best-of-N=102026.04 | 4.1 | 3.96 | 4.46 | 5 | 1 | 5 | |
| Our MethodModel=unsloth-Qwen2.5-7B, Strategy=BoN (10)2026.04 | 4.05 | 4.04 | 4.5 | 5 | 1.39 | 5 | |
| Research Agent2026.04 | 4.01 | 3.75 | 4.24 | 3.45 | 1.88 | 2.17 | |
| Research Agent2026.04 | 3.99 | 3.62 | 4.25 | 3.35 | 1 | 1.77 | |
| SFTBackbone=Qwen 2.5 14B, Mode=Supervised fine-tuning2026.04 | 3.95 | 3.93 | 4.22 | 2.68 | 2.55 | 1.43 | |
| GPT Researcher2026.04 | 3.93 | 4.01 | 4.35 | 3.55 | 2.1 | 3.33 | |
| GPT Researcher2026.04 | 3.84 | 4.17 | 4.33 | 3.3 | 2.31 | 3.19 | |
| AI Scientist V22026.04 | 3.8 | 4.28 | 4.22 | 2.59 | 3.52 | 3.51 | |
| AI Scientist2026.04 | 3.75 | 4.35 | 4.13 | 1.38 | 5 | 1 | |
| Base ModelModel=unsloth-Qwen2.5-7B2026.04 | 3.75 | 4.22 | 4.41 | 1.93 | 3.89 | 3.51 | |
| Base Model 14B 4-bitBackbone=Qwen 2.5 14B, Quantization=4-bit2026.04 | 3.71 | 4.31 | 4.32 | 3.71 | 4.34 | 4.32 | |
| AI Scientist2026.04 | 3.71 | 4.37 | 4.1 | 1 | 5 | 1 | |
| AI Scientist V22026.04 | 3.68 | 4.31 | 4.1 | 1 | 4.99 | 2.14 | |
| LDC2026.04 | 3.62 | 4.01 | 4.24 | 2.66 | 4.01 | 3.83 |