diff --git a/docs/source/getting_started/benchmarks.rst b/docs/source/getting_started/benchmarks.rst index 299611c9f..e25d45217 100644 --- a/docs/source/getting_started/benchmarks.rst +++ b/docs/source/getting_started/benchmarks.rst @@ -29,15 +29,15 @@ Evaluation settings are based on the following papers: :widths: 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20 "BarlowTwins", "Res50", "256", "100", "62.9", "84.3", "72.6", "90.9", "45.6", "73.9", "`link `_", "`link `_" - "BYOL", "Res50", "256", "100", "62.5", "85.0", "74.5", "92.0", "46.0", "74.8", "`link _`", "`link `_" + "BYOL", "Res50", "256", "100", "62.5", "85.0", "74.5", "92.0", "46.0", "74.8", "`link `_", "`link `_" "DINO", "Res50", "128", "100", "68.2", "87.9", "72.5", "90.8", "49.9", "78.7", "`link `_", "`link `_" - "MAE", "ViT-B/16", "256", "100", "46.0", "70.2", "81.3", "95.5", "11.2", "24.5", "`link _`", "`link `_" - "MoCoV2", "Res50", "256", "100", "61.5", "84.1", "74.3", "91.9", "41.8", "72.2", "`link _`", "`link `_" + "MAE", "ViT-B/16", "256", "100", "46.0", "70.2", "81.3", "95.5", "11.2", "24.5", "`link `_", "`link `_" + "MoCoV2", "Res50", "256", "100", "61.5", "84.1", "74.3", "91.9", "41.8", "72.2", "`link `_", "`link `_" "SimCLR*", "Res50", "256", "100", "63.2", "85.2", "73.9", "91.9", "44.8", "73.9", "`link `_", "`link `_" "SimCLR* + DCL", "Res50", "256", "100", "65.1", "86.2", "73.5", "91.7", "49.6", "77.5", "`link `_", "`link `_" "SimCLR* + DCLW", "Res50", "256", "100", "64.5", "86.0", "73.2", "91.5", "48.5", "76.8", "`link `_", "`link `_" "SwAV", "Res50", "256", "100", "67.2", "88.1", "75.4", "92.7", "49.5", "78.6", "`link `_", "`link `_" - "TiCo", "Res50", "256", "100", "49.7", "74.4", "72.7", "90.9", "26.6", "53.6", "`link _`", "`link `_" + "TiCo", "Res50", "256", "100", "49.7", "74.4", "72.7", "90.9", "26.6", "53.6", "`link `_", "`link `_" "VICReg", "Res50", "256", "100", "63.0", "85.4", "73.7", "91.9", "46.3", "75.2", "`link `_", "`link `_" *\*We use square root learning rate scaling instead of linear scaling as it yields better results for smaller batch sizes. See Appendix B.1 in the SimCLR paper.*