mirrors/pytorch-image-models
1
0
Fork 0
mirror of https://github.com/huggingface/pytorch-image-models.git synced 2025-06-03 15:01:08 +08:00
Code Issues Projects Releases Wiki Activity

Merge pull request #2202 from huggingface/mnv4_first_weights

Browse Source 
First set of MobileNetV4 weights trained in timm
This commit is contained in:
Ross Wightman 2024-06-11 23:06:22 -07:00 committed by GitHub
commit 5aa49d56bf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 24 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
timm/models/mobilenetv3.py
View File

@ -996,28 +996,28 @@ default_cfgs = generate_default_cfgs({
), ),
"lcnet_150.untrained": _cfg(), "lcnet_150.untrained": _cfg(),
'mobilenetv4_conv_small': _cfg( 'mobilenetv4_conv_small.e1200_r224_in1k': _cfg(
hf_hub_id='timm/',
test_input_size=(3, 256, 256), test_crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_conv_medium.e500_r224_in1k': _cfg(
hf_hub_id='timm/',
crop_pct=0.95, test_input_size=(3, 256, 256), test_crop_pct=1.0, interpolation='bicubic'),
'mobilenetv4_conv_medium.e500_r256_in1k': _cfg(
hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8),
crop_pct=0.95, test_input_size=(3, 320, 320), test_crop_pct=1.0, interpolation='bicubic'),
'mobilenetv4_conv_large.e500_r256_in1k': _cfg(
# hf_hub_id='timm/', # hf_hub_id='timm/',
interpolation='bicubic'), file='mobilenetv4_conv_large-r256.pth',
'mobilenetv4_conv_medium.r224': _cfg( input_size=(3, 256, 256), pool_size=(8, 8),
# hf_hub_id='timm/', crop_pct=0.95, test_input_size=(3, 320, 320), test_crop_pct=1.0, interpolation='bicubic'),
crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_conv_medium.r256': _cfg(
# hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_conv_large.r256': _cfg(
# hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_conv_large.r384': _cfg( 'mobilenetv4_conv_large.r384': _cfg(
# hf_hub_id='timm/', # hf_hub_id='timm/',
input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=0.95, interpolation='bicubic'), input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_hybrid_small': _cfg( 'mobilenetv4_hybrid_medium.e500_r224_in1k': _cfg(
# hf_hub_id='timm/', hf_hub_id='timm/',
interpolation='bicubic'), crop_pct=0.95, test_input_size=(3, 256, 256), test_crop_pct=1.0, interpolation='bicubic'),
'mobilenetv4_hybrid_medium.r224': _cfg(
# hf_hub_id='timm/',
crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_hybrid_medium.r256': _cfg( 'mobilenetv4_hybrid_medium.r256': _cfg(
# hf_hub_id='timm/', # hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'), input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'),
@ -1029,16 +1029,16 @@ default_cfgs = generate_default_cfgs({
input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=0.95, interpolation='bicubic'), input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=0.95, interpolation='bicubic'),
# experimental # experimental
'mobilenetv4_conv_aa_medium.r256': _cfg( 'mobilenetv4_conv_aa_medium.untrained': _cfg(
# hf_hub_id='timm/', # hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'), input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_conv_blur_medium.r256': _cfg( 'mobilenetv4_conv_blur_medium.e500_r224_in1k': _cfg(
# hf_hub_id='timm/', hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'), crop_pct=0.95, test_input_size=(3, 256, 256), test_crop_pct=1.0, interpolation='bicubic'),
'mobilenetv4_hybrid_medium_075': _cfg( 'mobilenetv4_hybrid_medium_075.untrained': _cfg(
# hf_hub_id='timm/', # hf_hub_id='timm/',
crop_pct=0.95, interpolation='bicubic'), crop_pct=0.95, interpolation='bicubic'),
'mobilenetv4_hybrid_large_075.r256': _cfg( 'mobilenetv4_hybrid_large_075.untrained': _cfg(
# hf_hub_id='timm/', # hf_hub_id='timm/',
input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'), input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=0.95, interpolation='bicubic'),
}) })
@ -1256,7 +1256,7 @@ def mobilenetv4_hybrid_medium_075(pretrained: bool = False, **kwargs) -> MobileN
@register_model @register_model
def mobilenetv4_hybrid_large_075(pretrained: bool = False, **kwargs) -> MobileNetV3: def mobilenetv4_hybrid_large_075(pretrained: bool = False, **kwargs) -> MobileNetV3:
""" MobileNet V4 Hybrid""" """ MobileNet V4 Hybrid"""
model = _gen_mobilenet_v4('mobilenetv4_hybrid_large', 0.75, pretrained=pretrained, **kwargs) model = _gen_mobilenet_v4('mobilenetv4_hybrid_large_075', 0.75, pretrained=pretrained, **kwargs)
return model return model