Multi-Property Molecular Optimization (plogP+DRD2) on ZINC 250k
54SR (%)MolMem
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MolMemBackbone=Qwen2.5-1.5B, Optimization Paradigm=Expert2026.04 | 54 | 48 | 13.78 | |
| Reinvent 4Backbone=Transformer, Optimization Paradigm=Novice2026.04 | 50 | 48 | 7.94 | |
| PEIT-LLMBackbone=Llama3.1-8B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 25 | 47 | 12.82 | |
| Direct RetrievalBackbone=N/A, Optimization Paradigm=Apprentice, Optimization Strategy=Retrieval-based2026.04 | 21.5 | 44 | 7.52 | |
| GeLLM3OBackbone=Llama3.1-8B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 16 | 58 | 7.41 | |
| SFT-onlyBackbone=Qwen2.5-1.5B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 11.5 | 62 | 7.16 | |
| Direct PromptBackbone=Qwen2.5-1.5B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 8 | 55 | 8.98 | |
| LlaSMolBackbone=Mistral-7B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 8 | 46 | 9.16 | |
| ChemLLMBackbone=ChemLLM-7B, Optimization Paradigm=Apprentice, Optimization Strategy=Task-Specific LLM2026.04 | 8 | 53 | 8.06 | |
| SFT-onlyBackbone=Qwen2.5-3B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 7.5 | 62 | 7.86 | |
| Direct PromptBackbone=Qwen2.5-3B, Optimization Paradigm=Apprentice, Optimization Strategy=Prompting & SFT2026.04 | 7 | 57 | 8.58 | |
| Graph-GABackbone=N/A, Optimization Paradigm=Novice2026.04 | 2.5 | 48 | 6.61 | |
| QMOBackbone=GRU, Optimization Paradigm=Novice2026.04 | 2 | 51 | 4.2 |