ResearchBenchmarksMCQ selection similarity on Uniform Synthetic DatasetFollow99.8Average SimilarityOracle86.2889.7993.396.81Mar 29, 2026Evaluation ResultsMethodMethodLinksAverage SimilarityOraclealpha=0alpha=02026.0399.8Oraclealpha=1alpha=12026.0397.3Oraclealpha=0.25alpha=0.252026.0395.7Oraclealpha=0.5alpha=0.52026.0394.4Oraclealpha=0.75alpha=0.752026.0394.4DQNalpha=0alpha=02026.0389.6DQNalpha=0.25alpha=0.252026.0387.7DQNalpha=1alpha=12026.0387.7DQNalpha=0.5alpha=0.52026.0387DQNalpha=0.75alpha=0.752026.0386.8