List your company

Mathematical Reasoning on GSM8k Tensor-2 (test)

49.2AccuracyAdaBack
-1.96811.31624.637.884Jun 22, 2025

Evaluation results

MethodLinks
AdaBackModel Size=3B2025.0649.2
SFT+RLModel Size=3B2025.0642.7
SFT+AdaBackModel Size=3B2025.0642.2
SFT+AdaBackModel Size=1B2025.0611.3
AdaBackModel Size=1B2025.068.5
SFT+RLModel Size=1B2025.066.9
Base+RLModel Size=1B2025.060
Base+RLModel Size=3B2025.060