def _dense(self, net, num_units, name=None, sparsity_technique='baseline'):
return sparse_fully_connected(
x=net,
units=num_units,
sparsity_technique=sparsity_technique,
kernel_regularizer=self._regularizer,
name=name)
def model(inputs, is_training):
"""Creation of the model graph."""
with tf.variable_scope(name, 'resnet_model'):
inputs = conv2d_fixed_padding(
inputs=inputs,
filters=int(64 * width),
kernel_size=7,
strides=2,
pruning_method=pruning_method
if prune_first_layer else 'baseline',
init_method=init_method if prune_first_layer else 'baseline',
data_format=data_format,
end_sparsity=end_sparsity,
weight_decay=weight_decay,
name='initial_conv')
inputs = tf.identity(inputs, 'initial_conv')
inputs = batch_norm_relu(inputs,
is_training,
data_format=data_format)
inputs = tf.layers.max_pooling2d(inputs=inputs,
pool_size=3,
strides=2,
padding='SAME',
data_format=data_format,
name='initial_max_pool')
inputs = tf.identity(inputs, 'initial_max_pool')
inputs = block_group(inputs=inputs,
filters=int(64 * width),
block_fn=block_fn,
blocks=num_blocks[0],
strides=1,
is_training=is_training,
name='block_group1',
pruning_method=pruning_method,
init_method=init_method,
data_format=data_format,
end_sparsity=end_sparsity,
weight_decay=weight_decay)
inputs = block_group(inputs=inputs,
filters=int(128 * width),
block_fn=block_fn,
blocks=num_blocks[1],
strides=2,
is_training=is_training,
name='block_group2',
pruning_method=pruning_method,
init_method=init_method,
data_format=data_format,
end_sparsity=end_sparsity,
weight_decay=weight_decay)
inputs = block_group(inputs=inputs,
filters=int(256 * width),
block_fn=block_fn,
blocks=num_blocks[2],
strides=2,
is_training=is_training,
name='block_group3',
pruning_method=pruning_method,
init_method=init_method,
data_format=data_format,
end_sparsity=end_sparsity,
weight_decay=weight_decay)
inputs = block_group(inputs=inputs,
filters=int(512 * width),
block_fn=block_fn,
blocks=num_blocks[3],
strides=2,
is_training=is_training,
name='block_group4',
pruning_method=pruning_method,
init_method=init_method,
data_format=data_format,
end_sparsity=end_sparsity,
weight_decay=weight_decay)
pool_size = (inputs.shape[1], inputs.shape[2])
inputs = tf.layers.average_pooling2d(inputs=inputs,
pool_size=pool_size,
strides=1,
padding='VALID',
data_format=data_format,
name='final_avg_pool')
inputs = tf.identity(inputs, 'final_avg_pool')
multiplier = 4 if block_fn is bottleneck_block_ else 1
fc_units = multiplier * int(512 * width)
inputs = tf.reshape(inputs, [-1, fc_units])
kernel_initializer = tf.random_normal_initializer(stddev=.01)
# If init_method==sparse and not pruning, skip.
if init_method != 'sparse' or prune_last_layer:
kernel_initializer = _pick_initializer(kernel_initializer,
init_method,
pruning_method,
end_sparsity)
kernel_regularizer = contrib_layers.l2_regularizer(weight_decay)
inputs = sparse_fully_connected(
x=inputs,
units=num_classes,
sparsity_technique=pruning_method
if prune_last_layer else 'baseline',
kernel_initializer=kernel_initializer,
kernel_regularizer=kernel_regularizer,
name='final_dense')
inputs = tf.identity(inputs, 'final_dense')
return inputs
def model(inputs, is_training):
"""Creation of the model graph."""
with tf.variable_scope(name, 'resnet_model'):
inputs = resnet_model.fixed_padding(inputs,
kernel_size=3,
data_format=data_format)
padding = 'VALID'
kernel_initializer = tf.variance_scaling_initializer()
kernel_regularizer = contrib_layers.l2_regularizer(weight_decay)
inputs = tf.layers.conv2d(inputs=inputs,
filters=_make_divisible(32 * width),
kernel_size=3,
strides=2,
padding=padding,
use_bias=False,
kernel_initializer=kernel_initializer,
kernel_regularizer=kernel_regularizer,
data_format=data_format,
name='initial_conv')
inputs = tf.identity(inputs, 'initial_conv')
inputs = resnet_model.batch_norm_relu(inputs,
is_training,
data_format=data_format)
mb_block = functools.partial(mbv1_block_,
is_training=is_training,
width=width,
pruning_method=pruning_method,
data_format=data_format,
weight_decay=weight_decay)
inputs = mb_block(inputs, filters=64, stride=1, block_id=0)
inputs = mb_block(inputs, filters=128, stride=2, block_id=1)
inputs = mb_block(inputs, filters=128, stride=1, block_id=2)
inputs = mb_block(inputs, filters=256, stride=2, block_id=3)
inputs = mb_block(inputs, filters=256, stride=1, block_id=4)
inputs = mb_block(inputs, filters=512, stride=2, block_id=5)
inputs = mb_block(inputs, filters=512, stride=1, block_id=6)
inputs = mb_block(inputs, filters=512, stride=1, block_id=7)
inputs = mb_block(inputs, filters=512, stride=1, block_id=8)
inputs = mb_block(inputs, filters=512, stride=1, block_id=9)
inputs = mb_block(inputs, filters=512, stride=1, block_id=10)
inputs = mb_block(inputs, filters=1024, stride=2, block_id=11)
inputs = mb_block(inputs, filters=1024, stride=1, block_id=12)
last_block_filters = _make_divisible(int(1024 * width), 8)
if data_format == 'channels_last':
pool_size = (inputs.shape[1], inputs.shape[2])
elif data_format == 'channels_first':
pool_size = (inputs.shape[2], inputs.shape[3])
inputs = tf.layers.average_pooling2d(inputs=inputs,
pool_size=pool_size,
strides=1,
padding='VALID',
data_format=data_format,
name='final_avg_pool')
inputs = tf.identity(inputs, 'final_avg_pool')
inputs = tf.reshape(inputs, [-1, last_block_filters])
kernel_initializer = tf.variance_scaling_initializer()
kernel_regularizer = contrib_layers.l2_regularizer(weight_decay)
if prune_last_layer:
inputs = sparse_fully_connected(
x=inputs,
units=num_classes,
sparsity_technique=pruning_method
if prune_last_layer else 'baseline',
kernel_initializer=kernel_initializer,
kernel_regularizer=kernel_regularizer,
name='final_dense')
else:
inputs = tf.layers.dense(inputs=inputs,
units=num_classes,
activation=None,
use_bias=True,
kernel_initializer=kernel_initializer,
kernel_regularizer=kernel_regularizer,
name='final_dense')
inputs = tf.identity(inputs, 'final_dense')
return inputs