ResearchBenchmarksEfficient Reasoning on LLM Reasoning GeneralEfficient Reasoning on LLM Reasoning GeneralFollow645,000Data SizeL1-2,920165,290333,500501,710Aug 13, 2025Evaluation resultsMethodMethodLinksData SizeTraining SamplesL1Type=RL & budget-forcingType=RL & budget-forcing2025.08645,000645,000DASTType=preference optimi...Type=preference optimization2025.08150,00020,600TrEffType=RLType=RL2025.0824,80024,800LCPOType=preference optimi...Type=preference optimization2025.0822,000800