List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 2025 (average@10)

1.4Average@10Qwen2.5-3B
-0.0560.3220.71.078May 26, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Qwen2.5-3BTraining Setting=Untuned2026.051.4
Qwen2.5-3BTraining Setting=RL fine-tuned with REVERSE-MATH-augmented samples2026.051.1
Llama3.2-3BTraining Setting=RL fine-tuned with duplicated GSM8K samples2026.050.7
Llama3.2-3BTraining Setting=RL fine-tuned with REVERSE-MATH-augmented samples2026.050.7
Qwen2.5-3BTraining Setting=RL fine-tuned with duplicated GSM8K samples2026.050.7
Llama3.1-8BTraining Setting=Untuned2026.050.4
Llama3.1-8BTraining Setting=RL fine-tuned with REVERSE-MATH-augmented samples2026.050.4
Mistral-7BTraining Setting=RL fine-tuned with REVERSE-MATH-augmented samples2026.050.4
Llama3.2-3BTraining Setting=Untuned2026.050
Llama3.1-8BTraining Setting=RL fine-tuned with duplicated GSM8K samples2026.050
Mistral-7BTraining Setting=Untuned2026.050
Mistral-7BTraining Setting=RL fine-tuned with duplicated GSM8K samples2026.050