List your company

Closed Loop RLHF on HH 20% noise

58.7Win RateURM
53.81255.08156.3557.619Jul 3, 2026

Evaluation results

MethodLinks
URMBase Model=Qwen-3-1.7B, RL Algorithm=GRPO, Number of generations=2, Epochs=32026.0758.7
α-URMBase Model=Qwen-3-1.7B, RL Algorithm=GRPO, Number of generations=2, Epochs=32026.0758.2
RMBase Model=Qwen-3-1.7B, RL Algorithm=GRPO, Number of generations=2, Epochs=32026.0754