ResearchDatasetsNVIDIA A100 GPUNVIDIA A100 GPUFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsModel Efficiency AnalysisNVIDIA A100 GPU21.5MACs (G)11Multi-Head Attention (MHA)NVIDIA A100 GPU207.2TFLOPS5Grouped-Query Attention (GQA)NVIDIA A100 GPU154.7TFLOPS2