Opponent modeling on CaSiNo (External)
53.84EMABOND
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BONDTraining strategy=Supervised SFT, Model scale=8B, Evaluation protocol=k-penalty2026.05 | 53.84 | 76.21 | 80.88 | |
| RoBERTa CD+CA+DNDEvaluation protocol=k-penalty2026.05 | 48.72 | 70.03 | 77.14 | |
| BERT CD+CA+DNDEvaluation protocol=k-penalty2026.05 | 44.22 | 69.21 | 76.03 | |
| 70B structured-CoT promptedModel scale=70B, Strategy=Prompted, Evaluation protocol=k-penalty2026.05 | 37.3 | 62.99 | 70.94 | |
| BoW-RankerEvaluation protocol=k-penalty2026.05 | 27.71 | 52.98 | 64.31 |