List your company

Mathematical Reasoning on HMMT Feb 25

73.3AVG@KSAPO + Distillation
25.87638.18850.562.812May 25, 2026

Evaluation results

MethodLinks
SAPO + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0573.3
CE + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0571.8
CEBase Model=Nanbeige4-3B-2511-Thinking2026.0570.6
SAPOBase Model=Nanbeige4-3B-2511-Thinking2026.0570.2
MSE + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0568.6
CEBase Model=Qwen3-4B-Thinking-25072026.0567.6
GRPOBase Model=Nanbeige4-3B-2511-Thinking2026.0567.3
CE + DistillationBase Model=Qwen3-4B-Thinking-25072026.0567.1
GRPO + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0567.1
PPOBase Model=Nanbeige4-3B-2511-Thinking2026.0565.4
BaseBase Model=Nanbeige4-3B-2511-Thinking2026.0563.9
MSE + DistillationBase Model=Qwen3-4B-Thinking-25072026.0562.7
SAPO + DistillationBase Model=Qwen3-4B-Thinking-25072026.0561.1
GRPO + DistillationBase Model=Qwen3-4B-Thinking-25072026.0561
SAPOBase Model=Qwen3-4B-Thinking-25072026.0560.3
GRPOBase Model=Qwen3-4B-Thinking-25072026.0557.8
PPOBase Model=Qwen3-4B-Thinking-25072026.0556.5
BaseBase Model=Qwen3-4B-Thinking-25072026.0556
MSEBase Model=Nanbeige4-3B-2511-Thinking2026.0531.8
MSEBase Model=Qwen3-4B-Thinking-25072026.0527.7