List your company
ResearchFormerly WizWand

Accuracy on AIME 24

28.54AccuracyDAPO
8.395213.625118.85524.0849Oct 1, 2025
Updated 8mo ago

Evaluation results

MethodLinks
DAPOAlgorithm=DAPO, Stage=Fully Trained, Base Model=Qwen3-8B2025.1028.54
AlphaRL-enhanced DAPOAlgorithm=DAPO, Stage=40%, Base Model=Qwen3-8B2025.1028.33
RLOOAlgorithm=RLOO, Stage=Fully Trained, Base Model=Qwen3-8B2025.1027.5
GRPOAlgorithm=GRPO, Stage=Fully Trained, Base Model=Qwen3-8B2025.1026.4
AlphaRL-enhanced GRPOAlgorithm=GRPO, Stage=40%, Base Model=Qwen3-8B2025.1022.25
AlphaRL-enhanced RLOOAlgorithm=RLOO, Stage=40%, Base Model=Qwen3-8B2025.1017.92
RLOOAlgorithm=RLOO, Stage=40%, Base Model=Qwen3-8B2025.1016.67
GRPOAlgorithm=GRPO, Stage=40%, Base Model=Qwen3-8B2025.1015.83
DAPOAlgorithm=DAPO, Stage=40%, Base Model=Qwen3-8B2025.1015.8
AlphaRL-enhanced DAPOAlgorithm=DAPO, Stage=10%, Base Model=Qwen3-8B2025.1015
DAPOAlgorithm=DAPO, Stage=10%, Base Model=Qwen3-8B2025.1012.5
AlphaRL-enhanced GRPOAlgorithm=GRPO, Stage=10%, Base Model=Qwen3-8B2025.1012.5
RLOOAlgorithm=RLOO, Stage=10%, Base Model=Qwen3-8B2025.1011.67
AlphaRL-enhanced RLOOAlgorithm=RLOO, Stage=10%, Base Model=Qwen3-8B2025.1011.67
GRPOAlgorithm=GRPO, Stage=10%, Base Model=Qwen3-8B2025.109.17