Davit revised (#4)

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

* Update davit.py

timm/models/davit.py

@@ -34,6 +34,7 @@ from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
 
 __all__ = ['DaViT']
 
+# modified nn.Sequential that includes a size tuple in the forward function
 class SequentialWithSize(nn.Sequential):
     def forward(self, x : Tensor, size: Tuple[int, int]):
         for module in self._modules.values():
@@ -171,12 +172,12 @@ class ChannelBlock(nn.Module):
                  ffn=True, cpe_act=False):
         super().__init__()
 
-        self.cpe = nn.ModuleList([ConvPosEnc(dim=dim, k=3, act=cpe_act),
+        self.cpe1 = ConvPosEnc(dim=dim, k=3, act=cpe_act)
-                                  ConvPosEnc(dim=dim, k=3, act=cpe_act)])
         self.ffn = ffn
         self.norm1 = norm_layer(dim)
         self.attn = ChannelAttention(dim, num_heads=num_heads, qkv_bias=qkv_bias)
         self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.cpe2 = ConvPosEnc(dim=dim, k=3, act=cpe_act)
         
         if self.ffn:
             self.norm2 = norm_layer(dim)
@@ -188,12 +189,12 @@ class ChannelBlock(nn.Module):
 
 
     def forward(self, x : Tensor, size: Tuple[int, int]):
-        x = self.cpe[0](x, size)
+        x = self.cpe1(x, size)
         cur = self.norm1(x)
         cur = self.attn(cur)
         x = x + self.drop_path(cur)
 
-        x = self.cpe[1](x, size)
+        x = self.cpe2(x, size)
         if self.ffn:
             x = x + self.drop_path(self.mlp(self.norm2(x)))
         return x, size
@@ -292,9 +293,8 @@ class SpatialBlock(nn.Module):
         self.num_heads = num_heads
         self.window_size = window_size
         self.mlp_ratio = mlp_ratio
-        self.cpe = nn.ModuleList([ConvPosEnc(dim=dim, k=3, act=cpe_act),
-                                  ConvPosEnc(dim=dim, k=3, act=cpe_act)])
         
+        self.cpe1 = ConvPosEnc(dim=dim, k=3, act=cpe_act)
         self.norm1 = norm_layer(dim)
         self.attn = WindowAttention(
             dim,
@@ -303,6 +303,7 @@ class SpatialBlock(nn.Module):
             qkv_bias=qkv_bias)
 
         self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.cpe2 = ConvPosEnc(dim=dim, k=3, act=cpe_act)
  
         if self.ffn:
             self.norm2 = norm_layer(dim)
@@ -318,7 +319,7 @@ class SpatialBlock(nn.Module):
         H, W = size
         B, L, C = x.shape
 
-        shortcut = self.cpe[0](x, size)
+        shortcut = self.cpe1(x, size)
         x = self.norm1(shortcut)
         x = x.view(B, H, W, C)
 
@@ -347,7 +348,7 @@ class SpatialBlock(nn.Module):
         x = x.view(B, H * W, C)
         x = shortcut + self.drop_path(x)
 
-        x = self.cpe[1](x, size)
+        x = self.cpe2(x, size)
         if self.ffn:
             x = x + self.drop_path(self.mlp(self.norm2(x)))
         return x, size
@@ -417,12 +418,10 @@ class DaViTStage(nn.Module):
         
     def forward(self, x : Tensor, size: Tuple[int, int]):
         x, size = self.patch_embed(x, size)
-        for block in self.blocks:
-            for layer in block:
         if self.grad_checkpointing and not torch.jit.is_scripting():
-                    x, size = checkpoint.checkpoint(layer, x, size)
+            x, size = checkpoint_seq(self.blocks, x, size)
         else:
-                    x, size = layer(x, size)
+            x, size = self.blocks(x, size)
         return x, size
 
 
@@ -490,8 +489,8 @@ class DaViT(nn.Module):
             stage = DaViTStage(
                 in_chans if stage_id == 0 else embed_dims[stage_id - 1],
                 embed_dims[stage_id],
-                depth = 1,
+                depth = depths[stage_id],
-                patch_size = patch_size,
+                patch_size = patch_size if stage_id == 0 else 2,
                 overlapped_patch = overlapped_patch,
                 attention_types = attention_types,
                 num_heads = num_heads[stage_id],
@@ -602,11 +601,14 @@ def checkpoint_filter_fn(state_dict, model):
     
     if 'state_dict' in state_dict:
         state_dict = state_dict['state_dict']
-
+    import re
     out_dict = {}
     for k, v in state_dict.items():
-        k = k.replace('main_blocks.', 'stages.stage_')
+        k = re.sub(r'patch_embeds.([0-9]+)', r'stages.\1.patch_embed', k)
+        k = re.sub(r'main_blocks.([0-9]+)', r'stages.\1.blocks', k)
         k = k.replace('head.', 'head.fc.')
+        k = k.replace('cpe.0', 'cpe1')
+        k = k.replace('cpe.1', 'cpe2')
         out_dict[k] = v
     return out_dict
     
@@ -642,7 +644,7 @@ def _cfg(url='', **kwargs): # not sure how this should be set up
         'num_classes': 1000, 'input_size': (3, 224, 224), 'pool_size': (7, 7),
         'crop_pct': 0.875, 'interpolation': 'bilinear',
         'mean': IMAGENET_DEFAULT_MEAN, 'std': IMAGENET_DEFAULT_STD,
-        'first_conv': 'patch_embeds.0.proj', 'classifier': 'head.fc',
+        'first_conv': 'stages.0.patch_embed.proj', 'classifier': 'head.fc',
         **kwargs
     }