Multi-Property Molecular Optimization (QED+plogP) on ZINC-250k
58Success Rate (SR)MolMem
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MolMemBackbone=Qwen2.5-1.5B, Optimization Paradigm=Expert2026.04 | 58 | 49 | 12.12 | |
| Direct RetrievalBackbone=N/A, Optimization Paradigm=Apprentice, Optimization Strategy=Retrieval-based2026.04 | 24 | 43 | 4.51 | |
| GeLLM3OBackbone=Llama3.1-8B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 19.5 | 60 | 2.89 | |
| SFT-onlyBackbone=Qwen2.5-3B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 18 | 59 | 2.89 | |
| LlaSMolBackbone=Mistral-7B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 15 | 49 | 3.58 | |
| SFT-onlyBackbone=Qwen2.5-1.5B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 14 | 60 | 3.46 | |
| Direct PromptBackbone=Qwen2.5-3B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 12.5 | 54 | 4.85 | |
| PEIT-LLMBackbone=Llama3.1-8B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 12.5 | 44 | 5.95 | |
| ChemLLMBackbone=ChemLLM-7B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 11.5 | 51 | 6.36 | |
| Direct PromptBackbone=Qwen2.5-1.5B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 10 | 58 | 3.49 | |
| Graph-GABackbone=N/A, Optimization Paradigm=Novice2026.04 | 8 | 50 | 4.11 | |
| Reinvent 4Backbone=Transformer, Optimization Paradigm=Novice2026.04 | 6 | 49 | 8.6 | |
| QMOBackbone=GRU, Optimization Paradigm=Novice2026.04 | 3 | 52 | 6.4 |