ResearchBenchmarksScience Question Answering on ARC (Relative Improvement Analysis)Follow14.03Average Relative ImprovementTBDF10.109211.127112.14513.1629Jan 29, 2026Evaluation ResultsMethodMethodLinksAverage Relative ImprovementInferior/Superior Case CountsTBDFFiltering Mode=FW-EDU,...Filtering Mode=FW-EDU, Models=Gemma 3 (270M, 1B, 4B)2026.0114.03—CBFiltering Mode=FW-EDU,...Filtering Mode=FW-EDU, Models=Gemma 3 (270M, 1B, 4B)2026.0111.01—TBDFFiltering Mode=General...Filtering Mode=General, Models=Gemma 3 (270M, 1B, 4B)2026.0110.26—