ResearchDatasetsA100FollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsGPU Inference ThroughputA100-SXM4-40GB5,081Throughput (tok/s)8LLM GenerationA100 80GB (inference)128Maximum Batch Size6Inference EfficiencyA100 80 GB GPU0.019Latency (s)5Model DiscoveryA1005vit_tiny Discovery Rate4