ResearchDatasetsQwen3 familyFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsTraining Efficiency ProfilingQwen3 family S=512 (train profiling)100.2Per-Step Latency (ms)43Training Efficiency ProfilingQwen3 family S=1024 (train profiling)100.6Per-Step Latency (ms)42