List your company

Mathematical Reasoning on AMC 2023 (Avg@32, Pass@32)

96.3Avg@32HTPO
21.73241.09160.4579.809Apr 12, 2026Apr 16, 2026Apr 20, 2026Apr 25, 2026Apr 29, 2026May 3, 2026May 8, 2026

Evaluation results

MethodLinks
HTPOBackbone=Qwen3-8B-Instruct2026.0596.3—
GSPOBackbone=Qwen3-8B-Instruct2026.0594.1—
SAPOBackbone=Qwen3-8B-Instruct2026.0594—
DAPOBackbone=Qwen3-8B-Instruct2026.0593.8—
80/20-RuleBackbone=Qwen3-8B-Instruct2026.0593.8—
BAPOBackbone=Qwen3-8B-Instruct2026.0593.3—
GRPO†Backbone=Qwen3-8B-Instruct2026.0592.2—
HTPOBackbone=Qwen3-8B-Base2026.0583—
GSPOBackbone=Qwen3-8B-Base2026.0582—
SAPOBackbone=Qwen3-8B-Base2026.0581.9—
SCOPE (Ours)Teacher Model=SkyWork-OR1-Math-7B, Student Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0480.997.2
BAPOBackbone=Qwen3-8B-Base2026.0577.8—
80/20-RuleBackbone=Qwen3-8B-Base2026.0577.2—
w/ OPDTeacher Model=SkyWork-OR1-Math-7B, Student Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0475.995
DAPOBackbone=Qwen3-8B-Base2026.0575.9—
w/ GRPOTeacher Model=SkyWork-OR1-Math-7B, Student Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0475.195
R1-Distill-Qwen-1.5BTeacher Model=SkyWork-OR1-Math-7B, Student Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0472.794.7
GRPO†Backbone=Qwen3-8B-Base2026.0570.2—
w/ KDTeacher Model=SkyWork-OR1-Math-7B, Student Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0469.196.3
SCOPE (Ours)Teacher Model=Qwen3-8B-Instruct, Student Model=Qwen3-1.7B-Base2026.0446.383
w/ OPDTeacher Model=Qwen3-8B-Instruct, Student Model=Qwen3-1.7B-Base2026.0443.180.1
w/ GRPOTeacher Model=Qwen3-8B-Instruct, Student Model=Qwen3-1.7B-Base2026.0433.160
w/ KDTeacher Model=Qwen3-8B-Instruct, Student Model=Qwen3-1.7B-Base2026.0429.866.1
Qwen3-1.7B-BaseTeacher Model=Qwen3-8B-Instruct, Student Model=Qwen3-1.7B-Base2026.0424.669.4