Poker Gameplay Performance on Leduc Hold'em
24.5NFSPQwen2.5-72B
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Qwen2.5-72BType=Vanilla2026.01 | 24.5 | -18 | -18 | -25 | -9.1 | |
| ToolPokerBackbone=Qwen2.5-7B2026.01 | 11.5 | 18 | 1 | -3 | 6.8 | |
| o4-miniType=Vanilla2026.01 | 11 | 20 | -33.5 | -8 | -2.6 | |
| BC-RIRLBackbone=Qwen2.5-7B2026.01 | -37 | -64.5 | -43.5 | -34 | -40.5 | |
| Qwen2.5-7BType=Vanilla2026.01 | -57.5 | -93 | -73 | -68.5 | -73 |