Add native PyTorch weights for MixNet-Medium with no SAME padding necessary. Remove unused block of code.

Author: rwightman

README.md

@@ -75,6 +75,7 @@ I've leveraged the training scripts in this repository to train a few of the mod
 | resnext50d_32x4d | 79.674 (20.326) | 94.868 (5.132) | 25.1M | bicubic |
 | resnext50_32x4d | 78.512 (21.488) | 94.042 (5.958) | 25M | bicubic |
 | resnet50 | 78.470 (21.530) | 94.266 (5.734) | 25.6M | bicubic |
+| mixnet_m | 77.256 (22.744) | 93.418 (6.582) | 5.01M | bicubic |
 | seresnext26_32x4d | 77.104 (22.896) | 93.316 (6.684) | 16.8M | bicubic |
 | efficientnet_b0 | 76.912 (23.088) | 93.210 (6.790) | 5.29M | bicubic |
 | resnet26d | 76.68 (23.32) | 93.166 (6.834) | 16M | bicubic |

timm/models/gen_efficientnet.py

@@ -103,7 +103,8 @@ def _cfg(url='', **kwargs):
         url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_efficientnet_b5-c6949ce9.pth',
         input_size=(3, 456, 456), pool_size=(15, 15), crop_pct=0.934),
     'mixnet_s': _cfg(url=''),
-    'mixnet_m': _cfg(url=''),
+    'mixnet_m': _cfg(
+        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/mixnet_m-4647fc68.pth'),
     'mixnet_l': _cfg(url=''),
     'tf_mixnet_s': _cfg(
         url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_mixnet_s-89d3354b.pth'),
@@ -271,13 +272,6 @@ def _decode_block_str(block_str, depth_multiplier=1.0):
     return [deepcopy(block_args) for _ in range(num_repeat)]
 
 
-def _decode_arch_args(string_list):
-    block_args = []
-    for block_str in string_list:
-        block_args.append(_decode_block_str(block_str))
-    return block_args
-
-
 def _decode_arch_def(arch_def, depth_multiplier=1.0):
     arch_args = []
     for stack_idx, block_strings in enumerate(arch_def):
@@ -1612,8 +1606,8 @@ def mixnet_m(pretrained=False, num_classes=1000, in_chans=3, **kwargs):
     model = _gen_mixnet_m(
         channel_multiplier=1.0, num_classes=num_classes, in_chans=in_chans, **kwargs)
     model.default_cfg = default_cfg
-    #if pretrained:
-    #    load_pretrained(model, default_cfg, num_classes, in_chans)
+    if pretrained:
+        load_pretrained(model, default_cfg, num_classes, in_chans)
     return model