ResearchDatasetsServeGenFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsMulti-modal to Audio Generation LatencyServeGen 16-GPU setup (real-world workload)21.3P50 Latency8Multi-modal to Text Generation LatencyServeGen 16-GPU setup (real-world workload)7.1Latency P508