update hello.py.

hello.py

@@ -23,18 +23,18 @@ import numpy as np
 
 def net():
 
-    model =  ViT(
+#     model =  ViT(
-    image_size = 600,
+#     image_size = 600,
-    patch_size = 30,
+#     patch_size = 30,
-    num_classes = 5,
+#     num_classes = 5,
-    dim = 1024,
+#     dim = 1024,
-    depth = 6,
+#     depth = 6,
-    max_tokens_per_depth = (256, 128, 64, 32, 16, 8), # a tuple that denotes the maximum number of tokens that any given layer should have. if the layer has greater than this amount, it will undergo adaptive token sampling
+#     max_tokens_per_depth = (256, 128, 64, 32, 16, 8), # a tuple that denotes the maximum number of tokens that any given layer should have. if the layer has greater than this amount, it will undergo adaptive token sampling
-    heads = 16,
+#     heads = 16,
-    mlp_dim = 2048,
+#     mlp_dim = 2048,
-    dropout = 0.1,
+#     dropout = 0.1,
-    emb_dropout = 0.1
+#     emb_dropout = 0.1
-)
+# )
 
 #     modelv = ViT(
 #     image_size = 600,
@@ -52,15 +52,15 @@ def net():
 # )
 
    
-#     model = NesT(
+    model = NesT(
-#     image_size = 600,
+    image_size = 600,
-#     patch_size = 30,
+    patch_size = 30,
-#     dim = 256,
+    dim = 256,
-#     heads = 16,
+    heads = 18,#16
-#     num_hierarchies = 3,        # number of hierarchies
+    num_hierarchies = 3,        # number of hierarchies
-#     block_repeats = (2, 2, 12),  # the number of transformer blocks at each heirarchy, starting from the bottom
+    block_repeats = (3, 3, 16),  # (2,2,12) the number of transformer blocks at each heirarchy, starting from the bottom
-#     num_classes = 5
+    num_classes = 5
-# )
+)
 
 #     model = CrossFormer( #图片尺寸要是7的倍数，如448
 #     num_classes = 5,                # number of output classes