List your company

Large Language Model Evaluation on LLM Core Evaluation Suite (test)

24.72GSM8KCDMA
24.636824.658424.6824.7016Nov 24, 2025

Evaluation results

MethodLinks
CDMAModel size=3.2B, Pretraining type=Continual pretraining, Schedule=Multi-domain curriculum + CDMA2025.1124.7238.1342.1248.9755.2577.9568.3974.7753.79
BaselineModel size=3.2B, Pretraining type=Continual pretraining, Schedule=Uniform + WSD LR2025.1124.6438.9140.8348.5351.8678.1367.3271.3852.7