More consistency in model arg/kwarg merge handling

Author: rwightman

timm/models/davit.py

@@ -648,41 +648,35 @@ def _cfg(url='', **kwargs):
 
 @register_model
 def davit_tiny(pretrained=False, **kwargs) -> DaVit:
-    model_kwargs = dict(
-        depths=(1, 1, 3, 1), embed_dims=(96, 192, 384, 768), num_heads=(3, 6, 12, 24), **kwargs)
-    return _create_davit('davit_tiny', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(1, 1, 3, 1), embed_dims=(96, 192, 384, 768), num_heads=(3, 6, 12, 24))
+    return _create_davit('davit_tiny', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def davit_small(pretrained=False, **kwargs) -> DaVit:
-    model_kwargs = dict(
-        depths=(1, 1, 9, 1), embed_dims=(96, 192, 384, 768), num_heads=(3, 6, 12, 24), **kwargs)
-    return _create_davit('davit_small', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(1, 1, 9, 1), embed_dims=(96, 192, 384, 768), num_heads=(3, 6, 12, 24))
+    return _create_davit('davit_small', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def davit_base(pretrained=False, **kwargs) -> DaVit:
-    model_kwargs = dict(
-        depths=(1, 1, 9, 1), embed_dims=(128, 256, 512, 1024), num_heads=(4, 8, 16, 32), **kwargs)
-    return _create_davit('davit_base', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(1, 1, 9, 1), embed_dims=(128, 256, 512, 1024), num_heads=(4, 8, 16, 32))
+    return _create_davit('davit_base', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def davit_large(pretrained=False, **kwargs) -> DaVit:
-    model_kwargs = dict(
-        depths=(1, 1, 9, 1), embed_dims=(192, 384, 768, 1536), num_heads=(6, 12, 24, 48), **kwargs)
-    return _create_davit('davit_large', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(1, 1, 9, 1), embed_dims=(192, 384, 768, 1536), num_heads=(6, 12, 24, 48))
+    return _create_davit('davit_large', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def davit_huge(pretrained=False, **kwargs) -> DaVit:
-    model_kwargs = dict(
-        depths=(1, 1, 9, 1), embed_dims=(256, 512, 1024, 2048), num_heads=(8, 16, 32, 64), **kwargs)
-    return _create_davit('davit_huge', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(1, 1, 9, 1), embed_dims=(256, 512, 1024, 2048), num_heads=(8, 16, 32, 64))
+    return _create_davit('davit_huge', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def davit_giant(pretrained=False, **kwargs) -> DaVit:
-    model_kwargs = dict(
-        depths=(1, 1, 12, 3), embed_dims=(384, 768, 1536, 3072), num_heads=(12, 24, 48, 96), **kwargs)
-    return _create_davit('davit_giant', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(1, 1, 12, 3), embed_dims=(384, 768, 1536, 3072), num_heads=(12, 24, 48, 96))
+    return _create_davit('davit_giant', pretrained=pretrained, **dict(model_args, **kwargs))

timm/models/densenet.py

@@ -361,8 +361,8 @@ def densenet121(pretrained=False, **kwargs) -> DenseNet:
     r"""Densenet-121 model from
     `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
     """
-    model = _create_densenet(
-        'densenet121', growth_rate=32, block_config=(6, 12, 24, 16), pretrained=pretrained, **kwargs)
+    model_args = dict(growth_rate=32, block_config=(6, 12, 24, 16))
+    model = _create_densenet('densenet121', pretrained=pretrained, **dict(model_args, **kwargs))
     return model
 
 
@@ -371,9 +371,8 @@ def densenetblur121d(pretrained=False, **kwargs) -> DenseNet:
     r"""Densenet-121 w/ blur-pooling & 3-layer 3x3 stem
     `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
     """
-    model = _create_densenet(
-        'densenetblur121d', growth_rate=32, block_config=(6, 12, 24, 16), pretrained=pretrained,
-        stem_type='deep', aa_layer=BlurPool2d, **kwargs)
+    model_args = dict(growth_rate=32, block_config=(6, 12, 24, 16), stem_type='deep', aa_layer=BlurPool2d)
+    model = _create_densenet('densenetblur121d', pretrained=pretrained, **dict(model_args, **kwargs))
     return model
 
 
@@ -382,8 +381,8 @@ def densenet169(pretrained=False, **kwargs) -> DenseNet:
     r"""Densenet-169 model from
     `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
     """
-    model = _create_densenet(
-        'densenet169', growth_rate=32, block_config=(6, 12, 32, 32), pretrained=pretrained, **kwargs)
+    model_args = dict(growth_rate=32, block_config=(6, 12, 32, 32))
+    model = _create_densenet('densenet169', pretrained=pretrained, **dict(model_args, **kwargs))
     return model
 
 
@@ -392,8 +391,8 @@ def densenet201(pretrained=False, **kwargs) -> DenseNet:
     r"""Densenet-201 model from
     `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
     """
-    model = _create_densenet(
-        'densenet201', growth_rate=32, block_config=(6, 12, 48, 32), pretrained=pretrained, **kwargs)
+    model_args = dict(growth_rate=32, block_config=(6, 12, 48, 32))
+    model = _create_densenet('densenet201', pretrained=pretrained, **dict(model_args, **kwargs))
     return model
 
 
@@ -402,8 +401,8 @@ def densenet161(pretrained=False, **kwargs) -> DenseNet:
     r"""Densenet-161 model from
     `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
     """
-    model = _create_densenet(
-        'densenet161', growth_rate=48, block_config=(6, 12, 36, 24), pretrained=pretrained, **kwargs)
+    model_args = dict(growth_rate=48, block_config=(6, 12, 36, 24))
+    model = _create_densenet('densenet161', pretrained=pretrained, **dict(model_args, **kwargs))
     return model
 
 
@@ -412,7 +411,7 @@ def densenet264d(pretrained=False, **kwargs) -> DenseNet:
     r"""Densenet-264 model from
     `"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>`
     """
-    model = _create_densenet(
-        'densenet264d', growth_rate=48, block_config=(6, 12, 64, 48), stem_type='deep', pretrained=pretrained, **kwargs)
+    model_args = dict(growth_rate=48, block_config=(6, 12, 64, 48), stem_type='deep')
+    model = _create_densenet('densenet264d', pretrained=pretrained, **dict(model_args, **kwargs))
     return model
 

timm/models/dpn.py

@@ -317,55 +317,55 @@ def _cfg(url='', **kwargs):
 
 @register_model
 def dpn48b(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         small=True, num_init_features=10, k_r=128, groups=32,
         b=True, k_sec=(3, 4, 6, 3), inc_sec=(16, 32, 32, 64), act_layer='silu')
-    return _create_dpn('dpn48b', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn48b', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def dpn68(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         small=True, num_init_features=10, k_r=128, groups=32,
         k_sec=(3, 4, 12, 3), inc_sec=(16, 32, 32, 64))
-    return _create_dpn('dpn68', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn68', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def dpn68b(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         small=True, num_init_features=10, k_r=128, groups=32,
         b=True, k_sec=(3, 4, 12, 3), inc_sec=(16, 32, 32, 64))
-    return _create_dpn('dpn68b', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn68b', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def dpn92(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         num_init_features=64, k_r=96, groups=32,
         k_sec=(3, 4, 20, 3), inc_sec=(16, 32, 24, 128))
-    return _create_dpn('dpn92', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn92', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def dpn98(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         num_init_features=96, k_r=160, groups=40,
         k_sec=(3, 6, 20, 3), inc_sec=(16, 32, 32, 128))
-    return _create_dpn('dpn98', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn98', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def dpn131(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         num_init_features=128, k_r=160, groups=40,
         k_sec=(4, 8, 28, 3), inc_sec=(16, 32, 32, 128))
-    return _create_dpn('dpn131', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn131', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def dpn107(pretrained=False, **kwargs) -> DPN:
-    model_kwargs = dict(
+    model_args = dict(
         num_init_features=128, k_r=200, groups=50,
         k_sec=(4, 8, 20, 3), inc_sec=(20, 64, 64, 128))
-    return _create_dpn('dpn107', pretrained=pretrained, **dict(model_kwargs, **kwargs))
+    return _create_dpn('dpn107', pretrained=pretrained, **dict(model_args, **kwargs))

timm/models/edgenext.py

@@ -528,8 +528,8 @@ def edgenext_xx_small(pretrained=False, **kwargs) -> EdgeNeXt:
     # No AA, Color Jitter=0.4, No Mixup & Cutmix, DropPath=0.0, BS=4096, lr=0.006, multi-scale-sampler
     # Jetson FPS=51.66 versus 47.67 for MobileViT_XXS
     # For A100: FPS @ BS=1: 212.13 & @ BS=256: 7042.06 versus FPS @ BS=1: 96.68 & @ BS=256: 4624.71 for MobileViT_XXS
-    model_kwargs = dict(depths=(2, 2, 6, 2), dims=(24, 48, 88, 168), heads=(4, 4, 4, 4), **kwargs)
-    return _create_edgenext('edgenext_xx_small', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(2, 2, 6, 2), dims=(24, 48, 88, 168), heads=(4, 4, 4, 4))
+    return _create_edgenext('edgenext_xx_small', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
@@ -539,8 +539,8 @@ def edgenext_x_small(pretrained=False, **kwargs) -> EdgeNeXt:
     # No AA, No Mixup & Cutmix, DropPath=0.0, BS=4096, lr=0.006, multi-scale-sampler
     # Jetson FPS=31.61 versus 28.49 for MobileViT_XS
     # For A100: FPS @ BS=1: 179.55 & @ BS=256: 4404.95 versus FPS @ BS=1: 94.55 & @ BS=256: 2361.53 for MobileViT_XS
-    model_kwargs = dict(depths=(3, 3, 9, 3), dims=(32, 64, 100, 192), heads=(4, 4, 4, 4), **kwargs)
-    return _create_edgenext('edgenext_x_small', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(3, 3, 9, 3), dims=(32, 64, 100, 192), heads=(4, 4, 4, 4))
+    return _create_edgenext('edgenext_x_small', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
@@ -550,8 +550,8 @@ def edgenext_small(pretrained=False, **kwargs) -> EdgeNeXt:
     # AA=True, No Mixup & Cutmix, DropPath=0.1, BS=4096, lr=0.006, multi-scale-sampler
     # Jetson FPS=20.47 versus 18.86 for MobileViT_S
     # For A100: FPS @ BS=1: 172.33 & @ BS=256: 3010.25 versus FPS @ BS=1: 93.84 & @ BS=256: 1785.92 for MobileViT_S
-    model_kwargs = dict(depths=(3, 3, 9, 3), dims=(48, 96, 160, 304),  **kwargs)
-    return _create_edgenext('edgenext_small', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=(3, 3, 9, 3), dims=(48, 96, 160, 304))
+    return _create_edgenext('edgenext_small', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
@@ -561,14 +561,14 @@ def edgenext_base(pretrained=False, **kwargs) -> EdgeNeXt:
     # AA=True, Mixup & Cutmix, DropPath=0.1, BS=4096, lr=0.006, multi-scale-sampler
     # Jetson FPS=xx.xx versus xx.xx for MobileViT_S
     # For A100: FPS @ BS=1: xxx.xx & @ BS=256: xxxx.xx
-    model_kwargs = dict(depths=[3, 3, 9, 3], dims=[80, 160, 288, 584], **kwargs)
-    return _create_edgenext('edgenext_base', pretrained=pretrained, **model_kwargs)
+    model_args = dict(depths=[3, 3, 9, 3], dims=[80, 160, 288, 584])
+    return _create_edgenext('edgenext_base', pretrained=pretrained, **dict(model_args, **kwargs))
 
 
 @register_model
 def edgenext_small_rw(pretrained=False, **kwargs) -> EdgeNeXt:
-    model_kwargs = dict(
+    model_args = dict(
         depths=(3, 3, 9, 3), dims=(48, 96, 192, 384),
-        downsample_block=True, conv_bias=False, stem_type='overlap', **kwargs)
-    return _create_edgenext('edgenext_small_rw', pretrained=pretrained, **model_kwargs)
+        downsample_block=True, conv_bias=False, stem_type='overlap')
+    return _create_edgenext('edgenext_small_rw', pretrained=pretrained, **dict(model_args, **kwargs))