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

Per-stage eca MV3 config, going to shelve these experiments for a while, not proving very productive

Browse Source
This commit is contained in:
Ross Wightman 2020-03-02 09:46:10 -08:00
parent ade1ba5fe3
commit 8bd08d5f64
2 changed files with 24 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
timm/models/efficientnet_builder.py
View File

@ -85,7 +85,8 @@ def _decode_block_str(block_str):
attn_layer = 'sev2'
attn_kwargs = dict(se_ratio=float(options['se']))
elif 'eca' in options:
attn_layer = 'eca'
attn_layer = 'ceca'
attn_kwargs = dict(kernel_size=int(options['eca']))
num_repeat = int(options['r'])
# each type of block has different valid arguments, fill accordingly
@ -378,7 +379,7 @@ class EfficientNetBuilder:
return stages
def _init_weight_goog(m, n='', fix_group_fanout=False):
def _init_weight_goog(m, n='', fix_group_fanout=True):
""" Weight initialization as per Tensorflow official implementations.
Args:

27
timm/models/mobilenetv3.py
View File

@ -389,23 +389,38 @@ def _gen_mobilenet_v3_eca(variant, channel_multiplier=1.0, pretrained=False, **k
else:
num_features = 1280
act_layer = HardSwish
# arch_def = [
# # stage 0, 112x112 in
# ['ds_r1_k3_s1_e1_c16_nre'], # relu
# # stage 1, 112x112 in
# ['ir_r1_k3_s2_e4_c24_nre', 'ir_r1_k3_s1_e3_c24_nre'], # relu
# # stage 2, 56x56 in
# ['ir_r3_k5_s2_e3_c40_nre'], # relu
# # stage 3, 28x28 in
# ['ir_r1_k3_s2_e6_c80', 'ir_r1_k3_s1_e2.5_c80', 'ir_r2_k3_s1_e2.3_c80'], # hard-swish
# # stage 4, 14x14in
# ['ir_r2_k3_s1_e6_c112'], # hard-swish
# # stage 5, 14x14in
# ['ir_r3_k5_s2_e6_c160'], # hard-swish
# # stage 6, 7x7 in
# ['cn_r1_k1_s1_c960'], # hard-swish
# ]
arch_def = [
# stage 0, 112x112 in
['ds_r1_k3_s1_e1_c16_nre'], # relu
# stage 1, 112x112 in
['ir_r1_k3_s2_e4_c24_nre', 'ir_r1_k3_s1_e3_c24_nre'], # relu
# stage 2, 56x56 in
['ir_r3_k5_s2_e3_c40_nre'], # relu
['ir_r3_k5_s2_e3_c40_eca3_nre'], # relu
# stage 3, 28x28 in
['ir_r1_k3_s2_e6_c80', 'ir_r1_k3_s1_e2.5_c80', 'ir_r2_k3_s1_e2.3_c80'], # hard-swish
['ir_r1_k3_s2_e6_c80_eca3', 'ir_r1_k3_s1_e2.5_c80_eca3', 'ir_r2_k3_s1_e2.3_c80_eca3'], # hard-swish
# stage 4, 14x14in
['ir_r2_k3_s1_e6_c112'], # hard-swish
['ir_r2_k3_s1_e6_c112_eca5'], # hard-swish
# stage 5, 14x14in
['ir_r3_k5_s2_e6_c160'], # hard-swish
['ir_r3_k5_s2_e6_c160_eca5'], # hard-swish
# stage 6, 7x7 in
['cn_r1_k1_s1_c960'], # hard-swish
]
model_kwargs = dict(
block_args=decode_arch_def(arch_def),
num_features=num_features,
@ -413,7 +428,7 @@ def _gen_mobilenet_v3_eca(variant, channel_multiplier=1.0, pretrained=False, **k
channel_multiplier=channel_multiplier,
norm_kwargs=resolve_bn_args(kwargs),
act_layer=act_layer,
attn_layer='ceca',
#attn_layer='ceca',
attn_kwargs=dict(gate_fn=hard_sigmoid),
**kwargs,
)