High level network definitions with pre-trained weights in TensorFlow
init
, load
, and save
(see examples).None.
reduce_max
and swish
.weights_regularizer
as a parameter of set_args
.None.
pretrained_initializer
with the latest NumPy.logits
to access a logits tensor directly.middles()
which is equivalent to get_middles()
.outputs()
which is equivalent to get_outputs()
.weights()
which is equivalent to get_weights()
.summary()
which is equivalent to print_summary()
.None.
variable_scope
and name_scope
(see #43).None.
None.
This is a small release that resolves installation issues on Windows.
pool1
in ResNets (the top-5 accuracies increased by 0.6-1.16%).None.
Note that the following descriptions include changes for the 0.3.3. Additionally, The 0.3.2 doesn't exist due to PyPI conflicts.
Darknet19
, MobileNet35v2
, ..., MobileNet140v2
, and PNASNetlarge
).YOLOv3VOC
and YOLOv3COCO
).local_flatten
and upsample
).None.
This is a small release that improves dependency importing and resolves PyPI conflicts.