DOFA: fix bug in patch embedding (#2256)

* DOFA: fix bug in patch embedding

* Update checkpoints

docs/api/weights/agnostic.csv

@@ -1,5 +1,5 @@
 Weight,Source,Citation,License,Spatial,Temporal,Spectral,m-bigearthnet,m-forestnet,m-brick-kiln,m-pv4ger,m-so2sat,m-eurosat,m-pv4ger-seg,m-nz-cattle,m-NeonTree,m-cashew-plant,m-SA-crop,m-chesapeake
-DOFABase16_Weights.DOFA_MAE,`link <https://github.com/zhu-xlab/DOFA>`__,`link <https://arxiv.org/abs/2403.15356>`__,CC-BY-4.0,implicit,-,explicit,63.8,45.3,94.7,96.9,52.1,92.2,94.7,81.6,58.6,48.3,31.3,65.4
-DOFALarge16_Weights.DOFA_MAE,`link <https://github.com/zhu-xlab/DOFA>`__,`link <https://arxiv.org/abs/2403.15356>`__,CC-BY-4.0,implicit,-,explicit,64.4,47.4,95.1,97.3,59.3,93.8,95.0,81.7,59.1,53.8,32.1,66.3
+DOFABase16_Weights.DOFA_MAE,`link <https://github.com/zhu-xlab/DOFA>`__,`link <https://arxiv.org/abs/2403.15356>`__,CC-BY-4.0,implicit,-,explicit,65.7,50.9,95.8,96.9,55.1,93.9,94.5,81.4,58.8,51.5,33.0,65.3
+DOFALarge16_Weights.DOFA_MAE,`link <https://github.com/zhu-xlab/DOFA>`__,`link <https://arxiv.org/abs/2403.15356>`__,CC-BY-4.0,implicit,-,explicit,67.5,54.6,96.9,97.3,60.1,97.1,95.0,81.8,59.4,56.9,32.1,66.3
 ResNet50_Weights.FMOW_RGB_GASSL,`link <https://github.com/sustainlab-group/geography-aware-ssl>`__,`link <https://arxiv.org/abs/2011.09980>`__,-,implicit,-,-,,,,,,,,,,,,
 ScaleMAE_ViTLarge16_Weights.FMOW_RGB_SCALEMAE,`link <https://github.com/bair-climate-initiative/scale-mae>`__,`link <https://arxiv.org/abs/2212.14532>`__,CC-BY-NC-4.0,explicit,-,-,,,,,,,,,,,

torchgeo/models/dofa.py

@@ -203,8 +203,10 @@ def forward(self, x: Tensor, wavelengths: Tensor) -> tuple[Tensor, Tensor]:
         weight, bias = self.weight_generator(waves)  # 3x3x3
 
         dynamic_weight = weight.view(
-            self.embed_dim, inplanes, self.kernel_size, self.kernel_size
-        )  # 3xoutdx16x16
+            inplanes, self.kernel_size, self.kernel_size, self.embed_dim
+        )
+        dynamic_weight = dynamic_weight.permute([3, 0, 1, 2])
+
         if bias is not None:
             bias = bias.view([self.embed_dim]) * self.scaler
 
@@ -384,7 +386,7 @@ class DOFABase16_Weights(WeightsEnum):  # type: ignore[misc]
     """
 
     DOFA_MAE = Weights(
-        url='https://hf.co/torchgeo/dofa/resolve/ade8745c5ec6eddfe15d8c03421e8cb8f21e66ff/dofa_base_patch16_224-7cc0f413.pth',
+        url='https://hf.co/torchgeo/dofa/resolve/b8db318b64a90b9e085ec04ba8851233c5893666/dofa_base_patch16_224-a0275954.pth',
         transforms=_dofa_transforms,
         meta={
             'dataset': 'SatlasPretrain, Five-Billion-Pixels, HySpecNet-11k',
@@ -403,7 +405,7 @@ class DOFALarge16_Weights(WeightsEnum):  # type: ignore[misc]
     """
 
     DOFA_MAE = Weights(
-        url='https://hf.co/torchgeo/dofa/resolve/ade8745c5ec6eddfe15d8c03421e8cb8f21e66ff/dofa_large_patch16_224-fbd47fa9.pth',
+        url='https://hf.co/torchgeo/dofa/resolve/b8db318b64a90b9e085ec04ba8851233c5893666/dofa_large_patch16_224-0ff904d3.pth',
         transforms=_dofa_transforms,
         meta={
             'dataset': 'SatlasPretrain, Five-Billion-Pixels, HySpecNet-11k',