Binary Classification on TDC (Therapeutic Data Commons)
82.3AMES ROC AUCRF
Evaluation results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RFAlgorithm=Random Forest, Features=RDKit descriptors, Trees=3002026.05 | 82.3 | 91.4 | 70.4 | 92.1 | 86.4 | 36 | 70.1 | 70.1 | 62.1 | 69.1 | 81.6 | 80.4 | 96.8 | 78.9 | 88.2 | 77.1 | |
| BOLEK (YN)Evaluation Protocol=YN, Rollouts=50, Temperature=0.62026.05 | 76.2 | 86.4 | 77.8 | 90.2 | 84.6 | 45 | 72.4 | 64.1 | 61.3 | 62.7 | 78.9 | 84 | 95.1 | 67.5 | 92.3 | 75.9 | |
| BOLEK (CoT)Evaluation Protocol=CoT, Rollouts=50, Temperature=0.62026.05 | 72.7 | 84.5 | 72.6 | 87.4 | 71.4 | 45.9 | 57.3 | 65 | 54 | 64.4 | 73.5 | 78.6 | 88.4 | 51.1 | 92.2 | 70.6 | |
| TxGemma-9B-ChatEvaluation Protocol=A/B token logit ratio2026.05 | 68 | 71.7 | 68.4 | 88.1 | 82.9 | 36.4 | 67.6 | 54.8 | 45.3 | 65.5 | 70.9 | 80.8 | 90.6 | 68.8 | 81.6 | 69.4 | |
| GPT-5.4Rollouts=52026.05 | 65.9 | 75.9 | 67.6 | 76.1 | 54.5 | 45.5 | 67.4 | 67.1 | 56.2 | 49.6 | 72.4 | 67.3 | 92.3 | 55.2 | 64.3 | 65.2 | |
| Qwen3-4B-InstructRollouts=50, Temperature=0.62026.05 | 50.8 | 73.7 | 49.4 | 71.7 | 54.3 | 25.6 | 61.6 | 31.4 | 55.8 | 53.4 | 68.3 | 65.9 | 64.2 | 45.4 | 59.7 | 55.4 |