ResearchBenchmarksText to Audio on ESNLI (test)Follow79.6AccuracySIFT41.32851.26461.271.136Apr 5, 2026Evaluation ResultsMethodMethodLinksAccuracySIFTRuntime (min)=51.3Runtime (min)=51.32026.0479.6BLURRuntime (min)=46.1Runtime (min)=46.12026.0477.4MuonRuntime (min)=47.9Runtime (min)=47.92026.0474.3AdamWRuntime (min)=44.6Runtime (min)=44.62026.0474.1POMERuntime (min)=24.7Runtime (min)=24.72026.0474.1Base2026.0442.8