mirrors
/
pytorch-image-models
mirror of https://github.com/huggingface/pytorch-image-models.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Add some so150m vit w/ sbb recipe weights, and a ese_vovnet57b model with RA4 recipe

pull/2415/head
Ross Wightman 2025-01-16 16:48:43 -08:00 committed by Ross Wightman
parent 9265d54a3a
commit 2a84d68d02
2 changed files with 24 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
timm/models/vision_transformer.py
View File

@ -2152,8 +2152,13 @@ default_cfgs = {
'vit_base_patch16_reg4_gap_256.untrained': _cfg(
input_size=(3, 256, 256)),
'vit_so150m_patch16_reg4_gap_256.untrained': _cfg(
input_size=(3, 256, 256)),
'vit_so150m_patch16_reg4_gap_384.sbb_e250_in12k_ft_in1k': _cfg(
hf_hub_id='timm/',
input_size=(3, 384, 384), crop_pct=1.0),
'vit_so150m_patch16_reg4_gap_256.sbb_e250_in12k': _cfg(
hf_hub_id='timm/',
num_classes=11821,
input_size=(3, 256, 256), crop_pct=0.95),
'vit_so150m_patch16_reg4_map_256.untrained': _cfg(
input_size=(3, 256, 256)),
@ -3482,6 +3487,17 @@ def vit_so150m_patch16_reg4_gap_256(pretrained: bool = False, **kwargs) -> Visio
return model
@register_model
def vit_so150m_patch16_reg4_gap_384(pretrained: bool = False, **kwargs) -> VisionTransformer:
model_args = dict(
patch_size=16, embed_dim=896, depth=18, num_heads=14, mlp_ratio=2.572,
class_token=False, reg_tokens=4, global_pool='avg', fc_norm=False,
)
model = _create_vision_transformer(
'vit_so150m_patch16_reg4_gap_384', pretrained=pretrained, **dict(model_args, **kwargs))
return model
@register_model
def vit_intern300m_patch14_448(pretrained: bool = False, **kwargs) -> VisionTransformer:
model_args = dict(

7
timm/models/vovnet.py
View File

@ -419,7 +419,12 @@ default_cfgs = generate_default_cfgs({
'ese_vovnet39b.ra_in1k': _cfg(
hf_hub_id='timm/',
test_input_size=(3, 288, 288), test_crop_pct=0.95),
'ese_vovnet57b.untrained': _cfg(url=''),
'ese_vovnet57b.ra4_e3600_r256_in1k': _cfg(
hf_hub_id='timm/',
mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5),
crop_pct=0.95, input_size=(3, 256, 256), pool_size=(8, 8),
test_input_size=(3, 320, 320), test_crop_pct=1.0
),
'ese_vovnet99b.untrained': _cfg(url=''),
'eca_vovnet39b.untrained': _cfg(url=''),
'ese_vovnet39b_evos.untrained': _cfg(url=''),