Game Performance Evaluation on Gin Rummy vs. prior self-play champion
51.4Win RateStacked recipe
Evaluation Results
| Method | Links | |
|---|---|---|
| Stacked recipeNotes=this work2026.07 | 51.4 | |
| TRPO baselineTraining steps=2M2026.07 | 50.5 | |
| PPO baselineTraining steps=2M2026.07 | 31 |