List your company
ResearchFormerly WizWand

Math Reasoning on AIME 2025 (avg@16 Accuracy)

11.25Accuracy (avg@16)GT-Reward
-0.452.58755.6258.6625Apr 28, 2026
Updated 6mo ago

Evaluation results

MethodLinks
GT-RewardBackbone=Qwen2.5-7B2026.0411.25
Self-CertaintyBackbone=Qwen2.5-7B2026.048.96
EntropyBackbone=Qwen2.5-7B2026.048.33
LLM-KDBackbone=Qwen2.5-7B2026.047.92
JURY-RLBackbone=Qwen2.5-7B2026.047.5
LLM-as-a-JudgeBackbone=Qwen2.5-7B2026.045.62
GT-RewardBackbone=Qwen3-1.7B-Base2026.045.21
JURY-RLBackbone=Qwen3-1.7B-Base2026.044.58
EntropyBackbone=Qwen3-1.7B-Base2026.044.17
Majority-VotingBackbone=Qwen2.5-7B2026.044.17
CoRewardBackbone=Qwen2.5-7B2026.043.96
LLM-KDBackbone=Qwen3-1.7B-Base2026.043.54
Self-CertaintyBackbone=Qwen3-1.7B-Base2026.043.12
CoRewardBackbone=Qwen3-1.7B-Base2026.043.12
Majority-VotingBackbone=Qwen3-1.7B-Base2026.042.08
LLM-as-a-JudgeBackbone=Qwen3-1.7B-Base2026.041.88
Before RLBackbone=Qwen2.5-7B2026.041.67
Before RLBackbone=Qwen3-1.7B-Base2026.041.25
JURY-RLBackbone=Llama-3.2-3B-Instruct2026.041.25
Self-CertaintyBackbone=Llama-3.2-3B-Instruct2026.040.62
LLM-as-a-JudgeBackbone=Llama-3.2-3B-Instruct2026.040.62
Before RLBackbone=Llama-3.2-3B-Instruct2026.040.42
LLM-KDBackbone=Llama-3.2-3B-Instruct2026.040.42
GT-RewardBackbone=Llama-3.2-3B-Instruct2026.040.21
EntropyBackbone=Llama-3.2-3B-Instruct2026.040.21
Majority-VotingBackbone=Llama-3.2-3B-Instruct2026.040
CoRewardBackbone=Llama-3.2-3B-Instruct2026.040