List your company

Scientific Reasoning on GPQA Diamond (Average Accuracy)

0.405Average AccuracyRePro
0.157480.221740.2860.35026Dec 1, 2025

Evaluation results

MethodLinks
ReProBackbone=Qwen3-4B-Base, Sample Count=Avg@4, Evaluation Domain=Out-of-domain, Base Method=GRPO2025.120.405
GRPOBackbone=Qwen3-4B-Base, Sample Count=Avg@4, Evaluation Domain=Out-of-domain2025.120.397
OriginalBackbone=Qwen3-4B-Base, Sample Count=Avg@4, Evaluation Domain=Out-of-domain2025.120.167