List your company

Mathematical Reasoning on GSM8K (Acc, Len, L-Acc)

96.6AccuracyPPO
72.26478.58284.991.218Jun 9, 2025

Evaluation results

MethodLinks
PPOBackbone=Qwen2.5-Math-7B-Instruct2025.0696.630591.5
Bingo-ABackbone=Qwen2.5-Math-7B-Instruct2025.0696.128391.5
Bingo-EBackbone=Qwen2.5-Math-7B-Instruct2025.069624192
BaseBackbone=Qwen2.5-Math-7B-Instruct2025.0695.833190.3
PPOBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0692.991887.5
Bingo-ABackbone=DeepSeek-R1-Distill-Qwen-7B2025.0692.337190.2
Bingo-EBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0691.836689.7
Bingo-ABackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.068756383.9
Bingo-EBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0686.734584.9
BaseBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0685.71,00180.3
PPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0685.41,31078.2
BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0673.22,02563.5