List your company
ResearchFormerly WizWand

Mathematical Reasoning on DeepMath (Peak Accuracy)

78Peak AccuracyDAPO
61.56865.83470.174.366May 26, 2026
Updated 5mo ago

Evaluation results

MethodLinks
DAPOModel Scale=14B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=24,5762026.0578
GRPOModel Scale=14B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=8,1922026.0577.4
PCModel Scale=14B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=12,2882026.0576.6
14B-PCModel Scale=14B, RL Algorithm=PC, Rollouts Trained=2,048, Rollouts Sampled=4,0962026.0575.6
14B-GRPOModel Scale=14B, RL Algorithm=GRPO, Rollouts Trained=2,048, Rollouts Sampled=2,0482026.0574.4
14B-DAPOModel Scale=14B, RL Algorithm=DAPO, Rollouts Trained=2,048, Rollouts Sampled=6,1442026.0573.4
PCModel Scale=1.5B, n (Rollout Budget)=128, Rollouts Trained=16,384, Rollouts Sampled=24,5762026.0572.6
PCModel Scale=8B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=12,2882026.0571.6
DAPOModel Scale=1.5B, n (Rollout Budget)=128, Rollouts Trained=16,384, Rollouts Sampled=49,1522026.0571
DAPOModel Scale=8B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=24,5762026.0570.2
8B-DAPOModel Scale=8B, RL Algorithm=DAPO, Rollouts Trained=2,048, Rollouts Sampled=6,1442026.0569.6
GRPOModel Scale=1.5B, n (Rollout Budget)=128, Rollouts Trained=16,384, Rollouts Sampled=16,3842026.0569.4
8B-PCModel Scale=8B, RL Algorithm=PC, Rollouts Trained=2,048, Rollouts Sampled=4,0962026.0568
4B-DAPOModel Scale=4B, RL Algorithm=DAPO, Rollouts Trained=2,048, Rollouts Sampled=6,1442026.0567
PCModel Scale=4B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=12,2882026.0567
8B-GRPOModel Scale=8B, RL Algorithm=GRPO, Rollouts Trained=2,048, Rollouts Sampled=2,0482026.0566.4
GRPOModel Scale=4B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=8,1922026.0566.2
4B-PCModel Scale=4B, RL Algorithm=PC, Rollouts Trained=2,048, Rollouts Sampled=4,0962026.0565.8
DAPOModel Scale=4B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=24,5762026.0565.6
1.5B-PCModel Scale=1.5B, RL Algorithm=PC, Rollouts Trained=2,048, Rollouts Sampled=4,0962026.0565.2
1.5B-DAPOModel Scale=1.5B, RL Algorithm=DAPO, Rollouts Trained=2,048, Rollouts Sampled=6,1442026.0565
GRPOModel Scale=8B, n (Rollout Budget)=64, Rollouts Trained=8,192, Rollouts Sampled=8,1922026.0564.2
1.5B-GRPOModel Scale=1.5B, RL Algorithm=GRPO, Rollouts Trained=2,048, Rollouts Sampled=2,0482026.0564
4B-GRPOModel Scale=4B, RL Algorithm=GRPO, Rollouts Trained=2,048, Rollouts Sampled=2,0482026.0562.2