Protein Sequence Generation on Target qA
0.822Composition ScoreProtGPT2 + RL (best run)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ProtGPT2 + RL (best run)Stage=RL, Criterion=lowest pool-mean JSD2026.06 | 0.822 | 0.0044 | 17.4 | |
| ProtGPT2 + RL (mean)Stage=RL, Seeds=302026.06 | 0.397 | 0.032 | 12.34 | |
| ProtGPT2 + Domain-Adapt. FTStage=Domain-Adaptive Fine-Tuning2026.06 | 0.126 | 0.059 | 9.17 | |
| Base ProtGPT2Stage=Base2026.06 | 0.007 | 0.247 | 3.7 |