Multi-objective Peptide Design on TfR Target Protein 1.0 (test)
10.614Binding AffinityOff-Policy RL (Fixed)
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Off-Policy RL (Fixed)Model length constraint=Fixed, Optimization strategy=Off-policy RL fine-tuning, Iterations=5002026.06 | 10.614 | 33.433 | 65.5 | 81 | 22.4 | 7.109 | |
| A2D2 w/o qualityModel length constraint=Any, Optimization strategy=AJD loss fine-tuned, Quality predictor configuration=None2026.06 | 10.241 | 41.267 | 79.2 | 83.1 | 9.6 | 6.102 | |
| A2D2 w/ both qualityModel length constraint=Any, Optimization strategy=AJD loss fine-tuned, Quality predictor configuration=both quality predictor optimization2026.06 | 10.19 | 48.6 | 77.6 | 83.1 | 11.8 | 6.338 | |
| Multi-Objective Guidance (Fixed)Model length constraint=Fixed, Optimization strategy=Inference-time multi-objective guidance, Guidance Model=PepTune2026.06 | 9.043 | — | 71.7 | 89.7 | 20.6 | 7.126 | |
| Pre-trained (Fixed)Model length constraint=Fixed, Optimization strategy=Unconditional2026.06 | 8.776 | 32.446 | 70.7 | 90.1 | 21.5 | 7.145 | |
| Pre-trained (Any)Model length constraint=Any, Optimization strategy=Unconditional2026.06 | 7.756 | 10.064 | 68.9 | 85.1 | 16.9 | 7.185 |