Mathematical Problem Solving on AMC 2023 (Peak Accuracy)
85Peak AccuracyPC
Evaluation Results
| Method | Links | |
|---|---|---|
| PCModel Scale=14B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=12,2882026.05 | 85 | |
| DAPOModel Scale=14B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=24,5762026.05 | 84.5 | |
| GRPOModel Scale=14B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=8,1922026.05 | 84.1 | |
| PCModel Scale=8B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=12,2882026.05 | 75.8 | |
| DAPOModel Scale=8B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=24,5762026.05 | 70.4 | |
| GRPOModel Scale=8B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=8,1922026.05 | 70.2 | |
| GRPOModel Scale=4B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=8,1922026.05 | 69.3 | |
| DAPOModel Scale=4B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=24,5762026.05 | 67 | |
| PCModel Scale=4B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=12,2882026.05 | 67 | |
| PCModel Scale=1.5B, n (Rollout Budget)=128, Rollouts Trained=16,384, Rollouts Sampled=24,5762026.05 | 65 | |
| DAPOModel Scale=1.5B, n (Rollout Budget)=128, Rollouts Trained=16,384, Rollouts Sampled=49,1522026.05 | 62.8 | |
| GRPOModel Scale=1.5B, n (Rollout Budget)=128, Rollouts Trained=16,384, Rollouts Sampled=16,3842026.05 | 60.9 |