update swin_doc

docs/zh_CN/models/SwinTransformer.md

@@ -60,7 +60,7 @@ Swin Transformer 是一种新的视觉 Transformer 网络，可以用作计算
 | SwinTransformer_base_patch4_window12_384                | 384       | 19.52                          | 64.56                          | 123.30                         |
 | SwinTransformer_base_patch4_window7_224<sup>[1]</sup>   | 224       | 13.53                          | 23.46                          | 39.13                          |
 | SwinTransformer_base_patch4_window12_384<sup>[1]</sup>  | 384       | 19.65                          | 64.72                          | 123.42                         |
-| SwinTransformer_large_patch4_window7_224<sup>[1]</sup>  | 224        15.74                          | 38.57                          | 71.49                          |
+| SwinTransformer_large_patch4_window7_224<sup>[1]</sup>  | 224       | 15.74                          | 38.57                          | 71.49                          |
 | SwinTransformer_large_patch4_window12_384<sup>[1]</sup> | 384       | 32.61                          | 116.59                         | 223.23                         |
 
 [1]：基于 ImageNet22k 数据集预训练，然后在 ImageNet1k 数据集迁移学习得到。