Molecular Property Optimization on macrocycle like
1.23Normalized Rewardπ(·|y*)
Evaluation Results
| Method | Links | |
|---|---|---|
| π(·|y*)2026.04 | 1.23 | |
| RLRL training budget=20002026.04 | 0.77 | |
| RLRL training budget=10002026.04 | 0.25 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=322026.04 | 0.21 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=642026.04 | 0.21 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=162026.04 | 0.18 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=82026.04 | 0.17 | |
| RLRL training budget=5002026.04 | 0.14 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=42026.04 | 0.13 | |
| Reward Weighted Classifier-Free Guidance (RCFG)Inference-time sampling budget |YS|=22026.04 | 0.08 |