ResearchBenchmarksStructured Document Translation on Inline markup dataset Zh→EnFollow39.92Content-BLEUPrompt31.932834.006436.0838.1536Dec 4, 2025Evaluation ResultsMethodMethodLinksContent-BLEUContent-COMETXML-ValidityXML-MatchXML-BLEUStrucAUCAverage ScorePrompt2025.1239.9283.3195.2683.6833.8367.0667.18FORMATRL2025.1234.7682.8395.7984.7434.7466.2866.52SFT2025.1232.2483.0695.2681.0533.0165.7765.07