add LeViT

README.md

@@ -79,6 +79,7 @@ pip install flowvision==0.1.0
             <li>gMLP</li>
             <li>ConvMixer</li>
             <li>ConvNeXt</li>
+            <li>LeViT</li>
             <li>RegionViT</li>
             <li>UniFormer</li>
             <li>VAN</li>

docs/source/changelog.md

@@ -16,7 +16,8 @@
 - Support [VAN](https://arxiv.org/abs/2202.09741) model and pretrained weight [#166](https://github.com/Oneflow-Inc/vision/pull/166)
 - Support [Dynamic convolution](https://arxiv.org/abs/1912.03458) module [#166](https://github.com/Oneflow-Inc/vision/pull/169)
 - Support transforms.RandomGrayscale method [#171](https://github.com/Oneflow-Inc/vision/pull/171)
-- Support [RegNet](https://arxiv.org/abs/2003.13678) model and pretrained weight [#166](https://github.com/Oneflow-Inc/vision/pull/170)
+- Support [RegNet](https://arxiv.org/abs/2003.13678) model and pretrained weight [#166](https://github.com/Oneflow-Inc/vision/pull/166)
+- Support [LeViT](https://arxiv.org/abs/2104.01136) model and pretrained weight [#177](https://github.com/Oneflow-Inc/vision/pull/177)
 
 **Bug Fixes**
 - Fix benchmark normalize mode error [#146](https://github.com/Oneflow-Inc/vision/pull/146)

docs/source/flowvision.models.rst

@@ -40,6 +40,7 @@ architectures for image classification:
 -  `gMLP`_
 -  `ConvMixer`_
 -  `ConvNeXt`_
+_  `LeViT`_
 -  `RegionViT`_
 -  `VAN`_
 
@@ -78,6 +79,7 @@ architectures for image classification:
 .. _ConvNeXt: https://arxiv.org/abs/2201.03545
 .. _RegionViT: https://arxiv.org/pdf/2106.02689.pdf
 .. _VAN: https://arxiv.org/pdf/2202.09741.pdf
+.. _LeViT: https://arxiv.org/pdf/2104.01136.pdf
 
 .. currentmodule:: flowvision.models
 
@@ -487,6 +489,17 @@ VAN
         van_large,
 
 
+LeViT
+--------
+.. automodule:: flowvision.models
+    :members:
+        levit_128s,
+        levit_128,
+        levit_192,
+        levit_256,
+        levit_384,
+
+
 Neural Style Transfer
 ==============
 

flowvision/models/__init__.py

@@ -30,6 +30,7 @@
 from .uniformer import *
 from .senet import *
 from .van import *
+from .levit import *
 
 from . import style_transfer
 from . import detection