Property optimization on TOMG-Bench
0.236QEDRePO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RePO2026.03 | 0.236 | 0.297 | 0.294 | |
| SFT2026.03 | 0.207 | 0.206 | 0.238 | |
| GRPOinitialization=SFT2026.03 | 0.193 | 0.183 | 0.225 | |
| Baseline2026.03 | 0.13 | 0.168 | 0.173 | |
| GRPO2026.03 | 0.123 | 0.305 | 0.188 | |
| BioT5-base2026.03 | 0.08 | 0.079 | 0.081 | |
| MolT5-large2026.03 | 0.055 | 0.043 | 0.048 |