def get_efficientnet_v2s_kwargs(channel_multiplier=1.0, depth_multiplier=1.0, **kwargs):
""" Creates an EfficientNet-V2s model
NOTE: this is a preliminary definition based on paper, awaiting official code release for details
and weights
Ref impl:
Paper: `EfficientNetV2: Smaller Models and Faster Training` - https://arxiv.org/abs/2104.00298
"""
arch_def = [
# FIXME it's not clear if the FusedMBConv layers have SE enabled for the Small variant,
# Table 4 suggests no. 23.94M params w/o, 23.98 with which is closer to 24M.
# ['er_r2_k3_s1_e1_c24_se0.25'],
# ['er_r4_k3_s2_e4_c48_se0.25'],
# ['er_r4_k3_s2_e4_c64_se0.25'],
['er_r2_k3_s1_e1_c24'],
['er_r4_k3_s2_e4_c48'],
['er_r4_k3_s2_e4_c64'],
['ir_r6_k3_s2_e4_c128_se0.25'],
['ir_r9_k3_s1_e6_c160_se0.25'],
['ir_r15_k3_s2_e6_c272_se0.25'],
]
model_kwargs = dict(
block_args=decode_arch_def(arch_def, depth_multiplier),
num_features=round_channels(1792, channel_multiplier, 8, None),
stem_size=24,
channel_multiplier=channel_multiplier,
norm_kwargs=resolve_bn_args(kwargs),
act_layer=resolve_act_layer(kwargs, 'silu'), # FIXME this is an assumption, paper does not mention
**kwargs,
)
return model_kwargs
def get_efficientnet_kwargs(channel_multiplier=1.0,
depth_multiplier=1.0,
drop_rate=0.2):
"""Creates an EfficientNet model.
Ref impl: https://github.com/tensorflow/tpu/blob/master/models/official/efficientnet/efficientnet_model.py
Paper: https://arxiv.org/abs/1905.11946
EfficientNet params
name: (channel_multiplier, depth_multiplier, resolution, dropout_rate)
'efficientnet-b0': (1.0, 1.0, 224, 0.2),
'efficientnet-b1': (1.0, 1.1, 240, 0.2),
'efficientnet-b2': (1.1, 1.2, 260, 0.3),
'efficientnet-b3': (1.2, 1.4, 300, 0.3),
'efficientnet-b4': (1.4, 1.8, 380, 0.4),
'efficientnet-b5': (1.6, 2.2, 456, 0.4),
'efficientnet-b6': (1.8, 2.6, 528, 0.5),
'efficientnet-b7': (2.0, 3.1, 600, 0.5),
'efficientnet-b8': (2.2, 3.6, 672, 0.5),
'efficientnet-l2': (4.3, 5.3, 800, 0.5),
Args:
channel_multiplier: multiplier to number of channels per layer
depth_multiplier: multiplier to number of repeats per stage
"""
arch_def = [
['ds_r1_k3_s1_e1_c16_se0.25'],
['ir_r2_k3_s2_e6_c24_se0.25'],
['ir_r2_k5_s2_e6_c40_se0.25'],
['ir_r3_k3_s2_e6_c80_se0.25'],
['ir_r3_k5_s1_e6_c112_se0.25'],
['ir_r4_k5_s2_e6_c192_se0.25'],
['ir_r1_k3_s1_e6_c320_se0.25'],
]
model_kwargs = dict(
block_args=decode_arch_def(arch_def, depth_multiplier),
num_features=round_channels(1280, channel_multiplier, 8, None),
stem_size=32,
channel_multiplier=channel_multiplier,
act_layer=Swish,
norm_kwargs={}, # TODO: check
drop_rate=drop_rate,
drop_path_rate=0.2,
)
return model_kwargs