Single-Property Molecular Optimization (JNK3) on ZINC-250k 200 lead molecules
98.5Success Rate (SR)MolMem
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MolMemBackbone=Qwen2.5-1.5B, Optimization Paradigm=Expert2026.04 | 98.5 | 47 | 8.87 | |
| ChemLLMBackbone=ChemLLM-7B, Optimization Paradigm=Apprentice, Method Category=Task-Specific LLM2026.04 | 44 | 65 | 2.48 | |
| PEIT-LLMBackbone=Llama3.1-8B, Optimization Paradigm=Apprentice, Method Category=Task-Specific LLM2026.04 | 41 | 50 | 2.51 | |
| Direct PromptBackbone=Qwen2.5-3B, Optimization Paradigm=Apprentice, Method Category=Prompting & SFT2026.04 | 31.5 | 64 | 1.79 | |
| Reinvent 4Backbone=Transformer, Optimization Paradigm=Novice2026.04 | 28 | 51 | 1.91 | |
| Direct RetrievalBackbone=N/A, Optimization Paradigm=Apprentice, Method Category=Retrieval-based2026.04 | 21 | 44 | 2.45 | |
| Direct PromptBackbone=Qwen2.5-1.5B, Optimization Paradigm=Apprentice, Method Category=Prompting & SFT2026.04 | 19.5 | 66 | 2.23 | |
| SFT-onlyBackbone=Qwen2.5-3B, Optimization Paradigm=Apprentice, Method Category=Prompting & SFT2026.04 | 10.5 | 64 | 0.81 | |
| MOLLEOBackbone=BioT5, Optimization Paradigm=Apprentice, Method Category=Task-Specific LLM2026.04 | 10 | 88 | 0.04 | |
| SFT-onlyBackbone=Qwen2.5-1.5B, Optimization Paradigm=Apprentice, Method Category=Prompting & SFT2026.04 | 8.5 | 63 | 0.72 | |
| GeLLM3OBackbone=Llama3.1-8B, Optimization Paradigm=Apprentice, Method Category=Task-Specific LLM2026.04 | 8.5 | 55 | 1.19 | |
| Graph-GABackbone=N/A, Optimization Paradigm=Novice2026.04 | 6.5 | 59 | 1.73 | |
| LlaSMolBackbone=Mistral-7B, Optimization Paradigm=Apprentice, Method Category=Task-Specific LLM2026.04 | 6 | 44 | 0.71 | |
| QMOBackbone=GRU, Optimization Paradigm=Novice2026.04 | 0.5 | 56 | 1.2 |