List your company

Mathematical Reasoning on GSM8K (avg@3 Accuracy)

95.9avg@3 AccuracyGemma-3-27B-IT
69.17276.11183.0589.989Jan 19, 2026

Evaluation results

MethodLinks
Gemma-3-27B-ITmax_completion_tokens=8K2026.0195.9
Qwen3-8B-Base + PASC-GRPOmax_completion_tokens=8K2026.0195.37
Gemma-3-12B-ITmax_completion_tokens=8K2026.0194.4
Qwen3-4B-Base + PASC-GRPOmax_completion_tokens=8K2026.0190.45
Qwen3-8B-Base + GRP-SFTmax_completion_tokens=8K2026.0187.72
LLaMA-3.1-8B-Instructmax_completion_tokens=8K2026.0185
Qwen3-4B-Base + GRP-SFTmax_completion_tokens=8K2026.0182.3
Qwen3-8B-Basemax_completion_tokens=8K2026.0173.01
Qwen3-4B-Basemax_completion_tokens=8K2026.0170.2