LLM Alignment on Gemma-3-4B
94.33Win RateNash Prox
Evaluation Results
| Method | Links | |
|---|---|---|
| Nash ProxOpponent=SFT2025.05 | 94.33 | |
| Online IPOOpponent=SFT2025.05 | 93.64 | |
| Online DPOOpponent=SFT2025.05 | 93.26 | |
| Nash ProxOpponent=Online IPO2025.05 | 56.27 | |
| Nash ProxOpponent=Online DPO2025.05 | 53.44 | |
| Online DPOOpponent=Online IPO2025.05 | 52.83 | |
| Online IPOOpponent=Online DPO2025.05 | 47.17 | |
| Online DPOOpponent=Nash Prox2025.05 | 46.56 | |
| Online IPOOpponent=Nash Prox2025.05 | 43.73 | |
| SFTOpponent=Online DPO2025.05 | 6.74 | |
| SFTOpponent=Online IPO2025.05 | 6.36 | |
| SFTOpponent=Nash Prox2025.05 | 5.67 |