List your company

Graduate-Level Science Question Answering on GPQA (P@1, P@4)

33.84P@1ICT
1.53769.923818.3126.6962Jun 18, 2026

Evaluation results

MethodLinks
ICTModel Size=Qwen2.5-7B2026.0633.8470.71
20-EntropyModel Size=Qwen2.5-7B2026.0631.9465.15
BaseModel Size=Qwen2.5-7B2026.0631.5767.68
GRPOModel Size=Qwen2.5-7B2026.0630.5464.14
STAPOModel Size=Qwen2.5-7B2026.0630.1864.14
STAPOModel Size=Qwen2.5-1.5B2026.0616.6529.34
ICTModel Size=Qwen2.5-1.5B2026.0616.429.71
BaseModel Size=Qwen2.5-1.5B2026.0614.0228.7
GRPOModel Size=Qwen2.5-1.5B2026.0613.8428.91
20-EntropyModel Size=Qwen2.5-1.5B2026.0611.7723.56
ICTModel Size=Qwen2.5-0.5B2026.065.9515.69
GRPOModel Size=Qwen2.5-0.5B2026.065.118.44
BaseModel Size=Qwen2.5-0.5B2026.064.5612.49
20-EntropyModel Size=Qwen2.5-0.5B2026.063.9810.75
STAPOModel Size=Qwen2.5-0.5B2026.062.788.37