ResearchBenchmarksStructured Document Translation on Inline markup dataset En→ZhFollow57.95Content-BLEUSFT54.663655.516856.3757.2232Dec 4, 2025Evaluation ResultsMethodMethodLinksContent-BLEUContent-COMETXML-ValidityXML-MatchXML-BLEUStrucAUCAverage ScoreSFT2025.1257.9586.1998.4289.4747.5163.1473.78FORMATRL2025.1257.786.2298.429047.6364.2974.04Prompt2025.1254.7985.8796.3284.7443.3356.6270.28