List your company

Mathematical Reasoning on AIME 2025 (avg@64, pass@64)

50.3Avg@64ASPO-Math-30B
17.0225.6634.342.94Oct 7, 2025

Evaluation results

MethodLinks
ASPO-Math-30BBackbone=Qwen3-30B-A3B2025.1050.380
GRPOBackbone=Qwen3-30B-A3B2025.1043.173.3
ASPO-Math-4BBackbone=Qwen3-4B2025.1040.970
ASPO-Math-8BBackbone=Qwen3-8B2025.1038.980
GRPOBackbone=Qwen3-4B2025.1035.570
ASPO-Math-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1035.170
GRPOBackbone=Qwen3-8B2025.1034.166.7
Nemotron-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1033.160
DeepScaleR-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.102963.3
GRPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1028.656.7
DeepSeek-R1-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1023.563.3
Qwen3-30B-A3BBackbone=Qwen3-30B-A3B2025.1019.956.7
Qwen3-8BBackbone=Qwen3-8B2025.1019.156.7
Qwen3-4BBackbone=Qwen3-4B2025.1018.363.3