ResearchDatasetsMixtralFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsTraining EfficiencyMixtral-8x22b-G8T8 Fine-grained28.8MFU5Training EfficiencyMixtral-8x22B Coarse-grained49.3MFU5Training Stability AnalysisMixtral 8x1B pre-training0Num Spikes2