List your company

Scientific Reasoning on GPQA (Acc., Maj. Vote)

67.42AccuracyMAS-Orchestra
24.759235.834646.9157.9854Jun 11, 2026

Evaluation results

MethodLinks
MAS-OrchestraContext=Training-from-scratch, Sampling Level=Trajectory2026.0667.4263.63
GRPOContext=Continued-training, Sampling Level=Trajectory2026.0667.164.65
Orch-RMContext=Continued-training, Sampling Level=Orchestration2026.0666.764.65
DPOContext=Continued-training2026.0666.4163.63
RFTContext=Continued-training, Sampling Level=Trajectory2026.066663.13
Orch-RMContext=Training-from-scratch, Sampling Level=Orchestration2026.0665.662.62
LLM-as-a-judgeContext=Continued-training, Sampling Level=Orchestration2026.0663.1960.1
Qwen2.5-7B-InstructContext=Base orchestrator2026.0626.417.17