Molecular Property Optimization on CNS Penetrant
57Normalized RewardRL
Evaluation Results
| Method | Links | |
|---|---|---|
| RLRL training budget=20002026.04 | 57 | |
| π(·|y*)2026.04 | 52 | |
| RLRL training budget=10002026.04 | 50 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=642026.04 | 48 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=322026.04 | 47 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=162026.04 | 43 | |
| RLRL training budget=5002026.04 | 42 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=82026.04 | 33 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=42026.04 | 23 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=22026.04 | 4 |