List your company

Mathematical Reasoning on AMC 2023 (avg@64, pass@64)

94.9Avg@64ASPO-Math-30B
66.61273.95681.388.644Oct 7, 2025

Evaluation results

MethodLinks
ASPO-Math-30BBackbone=Qwen3-30B-A3B2025.1094.997.5
GRPOBackbone=Qwen3-30B-A3B2025.1091.997.5
ASPO-Math-8BBackbone=Qwen3-8B2025.1089.197.5
ASPO-Math-4BBackbone=Qwen3-4B2025.1087.497.5
ASPO-Math-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1087.295
Nemotron-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1086.197.5
GRPOBackbone=Qwen3-4B2025.1084.397.5
GRPOBackbone=Qwen3-8B2025.1084.195
DeepScaleR-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1081.3100
GRPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1080.397.5
Qwen3-30B-A3BBackbone=Qwen3-30B-A3B2025.1074.3100
DeepSeek-R1-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1070.7100
Qwen3-8BBackbone=Qwen3-8B2025.1068.997.5
Qwen3-4BBackbone=Qwen3-4B2025.1067.795