ResearchBenchmarksClosed Loop RLHF on HH 20% noiseFollow58.7Win RateURM53.81255.08156.3557.619Jul 3, 2026Evaluation ResultsMethodMethodLinksWin RateURMBase Model=Qwen-3-1.7B...Base Model=Qwen-3-1.7B, RL Algorithm=GRPO, Number of generations=2, Epochs=32026.0758.7α-URMBase Model=Qwen-3-1.7B...Base Model=Qwen-3-1.7B, RL Algorithm=GRPO, Number of generations=2, Epochs=32026.0758.2RMBase Model=Qwen-3-1.7B...Base Model=Qwen-3-1.7B, RL Algorithm=GRPO, Number of generations=2, Epochs=32026.0754