ResearchBenchmarksImage Classification on ImageNet-1K (val) (Top-1/Top-5 Accuracy, Compression, and Throughput)For AI assistantsMarkdownContent indexImage Classification on ImageNet-1K (val) (Top-1/Top-5 Accuracy, Compression, and Throughput)Follow84.7Top-1 AccuracySwin-MoE84.49284.54684.684.654Nov 13, 2024Evaluation resultsMethodMethodLinksTop-1 AccuracyTop-5 AccuracyCompression RateThroughput (samples/s)Swin-MoEArchitecture=Swin-MoE,...Architecture=Swin-MoE, Protocol=Fine-tuning, Variant=Origin2024.1184.797—184.3LSH-MoEArchitecture=Swin-MoE,...Architecture=Swin-MoE, Protocol=Fine-tuning, Variant=Ours2024.1184.597.111.7236.6