def __init__(self, data_shape, augment_size, num_steps,
mid_channels, num_context, num_blocks, dropout, num_mixtures, checkerboard=True, tuple_flip=True, coupling_network="transformer"):
context_in = data_shape[0] * 2 if checkerboard else data_shape[0]
layers = []
if checkerboard: layers.append(Checkerboard(concat_dim=1))
layers += [Conv2d(context_in, mid_channels // 2, kernel_size=3, stride=1),
nn.Conv2d(mid_channels // 2, mid_channels, kernel_size=2, stride=2, padding=0),
GatedConvNet(channels=mid_channels, num_blocks=2, dropout=0.0),
Conv2dZeros(in_channels=mid_channels, out_channels=num_context)]
context_net = nn.Sequential(*layers)
# layer transformations of the augment flow
transforms = []
sample_shape = (augment_size * 4, data_shape[1] // 2, data_shape[2] // 2)
for i in range(num_steps):
flip = (i % 2 == 0) if tuple_flip else False
transforms.append(ActNormBijection2d(sample_shape[0]))
transforms.extend([Conv1x1(sample_shape[0])])
if coupling_network in ["conv", "densenet"]:
# just included for debugging
transforms.append(
ConditionalCoupling(in_channels=sample_shape[0],
num_context=num_context,
num_blocks=num_blocks,
mid_channels=mid_channels,
depth=1,
dropout=dropout,
gated_conv=False,
coupling_network=coupling_network,
checkerboard=checkerboard,
flip=flip))
elif coupling_network == "transformer":
transforms.append(
ConditionalMixtureCoupling(in_channels=sample_shape[0],
num_context=num_context,
mid_channels=mid_channels,
num_mixtures=num_mixtures,
num_blocks=num_blocks,
dropout=dropout,
use_attn=False,
checkerboard=checkerboard,
flip=flip))
else:
raise ValueError(f"Unknown network type {coupling_network}")
# Final shuffle of channels, squeeze and sigmoid
transforms.extend([Conv1x1(sample_shape[0]),
Unsqueeze2d(),
Sigmoid()])
super(AugmentFlow, self).__init__(base_dist=StandardNormal(sample_shape), # ConvNormal2d(sample_shape),
transforms=transforms,
context_init=context_net)
def test_bijection_is_well_behaved(self):
batch_size = 10
shape = [3, 32, 32]
x = torch.rand(batch_size, *shape)
bijections = [
Conv1x1(3, orthogonal_init=True),
Conv1x1(3, orthogonal_init=False),
]
self.eps = 1e-4
for bijection in bijections:
with self.subTest(bijection=bijection):
self.assert_bijection_is_well_behaved(bijection,
x,
z_shape=(batch_size,
*shape))
def __init__(self, data_shape, num_bits, num_steps, num_context,
num_blocks, mid_channels, depth, growth, dropout, gated_conv):
context_net = nn.Sequential(LambdaLayer(lambda x: 2*x.float()/(2**num_bits-1)-1),
DenseBlock(in_channels=data_shape[0],
out_channels=mid_channels,
depth=4,
growth=16,
dropout=dropout,
gated_conv=gated_conv,
zero_init=False),
nn.Conv2d(mid_channels, mid_channels, kernel_size=2, stride=2, padding=0),
DenseBlock(in_channels=mid_channels,
out_channels=num_context,
depth=4,
growth=16,
dropout=dropout,
gated_conv=gated_conv,
zero_init=False))
transforms = []
sample_shape = (data_shape[0] * 4, data_shape[1] // 2, data_shape[2] // 2)
for i in range(num_steps):
transforms.extend([
Conv1x1(sample_shape[0]),
ConditionalCoupling(in_channels=sample_shape[0],
num_context=num_context,
num_blocks=num_blocks,
mid_channels=mid_channels,
depth=depth,
growth=growth,
dropout=dropout,
gated_conv=gated_conv)
])
# Final shuffle of channels, squeeze and sigmoid
transforms.extend([Conv1x1(sample_shape[0]),
Unsqueeze2d(),
Sigmoid()
])
super(DequantizationFlow, self).__init__(base_dist=ConvNormal2d(sample_shape),
transforms=transforms,
context_init=context_net)
num_blocks=1,
mid_channels=64,
depth=8,
growth=16,
dropout=0.0,
gated_conv=True,
zero_init=True), ElementwiseParams2d(2))
model = Flow(base_dist=StandardNormal((24, 8, 8)),
transforms=[
UniformDequantization(num_bits=8),
Augment(StandardUniform((3, 32, 32)), x_size=3),
AffineCouplingBijection(net(6)),
ActNormBijection2d(6),
Conv1x1(6),
AffineCouplingBijection(net(6)),
ActNormBijection2d(6),
Conv1x1(6),
AffineCouplingBijection(net(6)),
ActNormBijection2d(6),
Conv1x1(6),
AffineCouplingBijection(net(6)),
ActNormBijection2d(6),
Conv1x1(6),
Squeeze2d(),
Slice(StandardNormal((12, 16, 16)), num_keep=12),
AffineCouplingBijection(net(12)),
ActNormBijection2d(12),
Conv1x1(12),
AffineCouplingBijection(net(12)),
dropout=0.0,
gated_conv=True,
zero_init=True),
ElementwiseParams2d(2))
#model = NDPFlow(base_dist=[StandardNormal((16,7,7)), StandardNormal((latent_size,))],
model = NDPFlow(base_dist=[None, StandardNormal((latent_size,))],
transforms=[
UniformDequantization(num_bits=8),
ActNormBijection2d(1),
ScalarAffineBijection(scale=2.0),
ScalarAffineBijection(shift=-0.5),
Squeeze2d(),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4)),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4)),
Squeeze2d(),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16)),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16)),
VAE(encoder=encoder, decoder=decoder)
]).to(device)
print(model)
###########
## Optim ##
###########
optimizer = Adam(model.parameters(), lr=1e-3)
def __init__(self, data_shape, num_bits,
base_distribution, num_scales, num_steps, actnorm,
vae_hidden_units,
coupling_network,
dequant, dequant_steps, dequant_context,
coupling_blocks, coupling_channels, coupling_dropout,
coupling_gated_conv=None, coupling_depth=None, coupling_mixtures=None):
assert len(base_distribution) == 1, "Only a single base distribution is supported"
transforms = []
current_shape = data_shape
if num_steps == 0: num_scales = 0
if dequant == 'uniform' or num_steps == 0 or num_scales == 0:
# no bijective flows defaults to only using uniform dequantization
transforms.append(UniformDequantization(num_bits=num_bits))
elif dequant == 'flow':
dequantize_flow = DequantizationFlow(data_shape=data_shape,
num_bits=num_bits,
num_steps=dequant_steps,
coupling_network=coupling_network,
num_context=dequant_context,
num_blocks=coupling_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
dropout=coupling_dropout,
gated_conv=coupling_gated_conv,
num_mixtures=coupling_mixtures)
transforms.append(VariationalDequantization(encoder=dequantize_flow, num_bits=num_bits))
# Change range from [0,1]^D to [-0.5, 0.5]^D
transforms.append(ScalarAffineBijection(shift=-0.5))
for scale in range(num_scales):
# squeeze to exchange height and width for more channels
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4,
current_shape[1] // 2,
current_shape[2] // 2)
# Dimension preserving components
for step in range(num_steps):
if actnorm: transforms.append(ActNormBijection2d(current_shape[0]))
transforms.append(Conv1x1(current_shape[0]))
if coupling_network == "conv":
transforms.append(
Coupling(in_channels=current_shape[0],
num_blocks=coupling_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
dropout=coupling_dropout,
gated_conv=coupling_gated_conv,
coupling_network=coupling_network))
else:
transforms.append(
MixtureCoupling(in_channels=current_shape[0],
mid_channels=coupling_channels,
num_mixtures=coupling_mixtures,
num_blocks=coupling_blocks,
dropout=coupling_dropout))
# Non-dimension preserving flows: reduce the dimensionality of data by 2 (channel-wise)
if actnorm: transforms.append(ActNormBijection2d(current_shape[0]))
assert current_shape[0] % 2 == 0, f"Current shape {current_shape[1]}x{current_shape[2]} must be divisible by two"
latent_size = (current_shape[0] * current_shape[1] * current_shape[2]) // 2
encoder = ConditionalNormal(
ConvEncoderNet(in_channels=current_shape[0],
out_channels=latent_size,
mid_channels=vae_hidden_units,
max_pool=True, batch_norm=True),
split_dim=1)
decoder = ConditionalNormal(
ConvDecoderNet(in_channels=latent_size,
out_shape=(current_shape[0] * 2, current_shape[1], current_shape[2]),
mid_channels=list(reversed(vae_hidden_units)),
batch_norm=True,
in_lambda=lambda x: x.view(x.shape[0], x.shape[1], 1, 1)),
split_dim=1)
transforms.append(VAE(encoder=encoder, decoder=decoder))
current_shape = (current_shape[0] // 2,
current_shape[1],
current_shape[2])
if scale < num_scales - 1:
# reshape latent sample to have height and width
transforms.append(Reshape(input_shape=(latent_size,), output_shape=current_shape))
# Base distribution for dimension preserving portion of flow
if base_distribution == "n":
base_dist = StandardNormal((latent_size,))
elif base_distribution == "c":
base_dist = ConvNormal2d((latent_size,))
elif base_distribution == "u":
base_dist = StandardUniform((latent_size,))
else:
raise ValueError("Base distribution must be one of n=Normal, u=Uniform, or c=ConvNormal")
# for reference save the shape output by the bijective flow
self.latent_size = latent_size
self.flow_shape = current_shape
super(MultilevelCompressiveFlow, self).__init__(base_dist=[None, base_dist], transforms=transforms)
def __init__(self, data_shape, cond_shape, num_bits, num_scales, num_steps,
actnorm, pooling, dequant, dequant_steps, dequant_context,
densenet_blocks, densenet_channels, densenet_depth,
densenet_growth, dropout, gated_conv, init_context):
transforms = []
current_shape = data_shape
if dequant == 'uniform':
transforms.append(UniformDequantization(num_bits=num_bits))
elif dequant == 'flow':
dequantize_flow = DequantizationFlow(
data_shape=data_shape,
num_bits=num_bits,
num_steps=dequant_steps,
num_context=dequant_context,
num_blocks=densenet_blocks,
mid_channels=densenet_channels,
depth=densenet_depth,
dropout=dropout,
gated_conv=gated_conv)
transforms.append(
VariationalDequantization(encoder=dequantize_flow,
num_bits=num_bits))
# Change range from [0,1]^D to [-0.5, 0.5]^D
transforms.append(ScalarAffineBijection(shift=-0.5))
# Initial squeeze
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
# Pooling flows
for scale in range(num_scales):
for step in range(num_steps):
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
transforms.extend([
Conv1x1(num_channels=current_shape[0]),
#ConditionalConv1x1(cond_shape=cond_shape, num_channels=current_shape[0]), # for conditional images!
ConditionalCoupling(in_channels=current_shape[0],
num_context=cond_shape[0],
num_blocks=densenet_blocks,
mid_channels=densenet_channels,
depth=densenet_depth,
dropout=dropout,
gated_conv=gated_conv)
])
if scale < num_scales - 1:
if pooling == 'none':
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4,
current_shape[1] // 2,
current_shape[2] // 2)
else:
if pooling == 'slice':
noise_shape = (current_shape[0] * 2,
current_shape[1] // 2,
current_shape[2] // 2)
transforms.append(Squeeze2d())
transforms.append(
Slice(StandardNormal(noise_shape),
num_keep=current_shape[0] * 2,
dim=1))
current_shape = (current_shape[0] * 2,
current_shape[1] // 2,
current_shape[2] // 2)
elif pooling == 'max':
noise_shape = (current_shape[0] * 3,
current_shape[1] // 2,
current_shape[2] // 2)
decoder = StandardHalfNormal(noise_shape)
transforms.append(
SimpleMaxPoolSurjection2d(decoder=decoder))
current_shape = (current_shape[0],
current_shape[1] // 2,
current_shape[2] // 2)
else:
raise ValueError(
"pooling argument must be either slice, max or none"
)
else:
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
# for reference save the shape output by the bijective flow
self.flow_shape = current_shape
super(CondPoolFlow,
self).__init__(base_dist=ConvNormal2d(current_shape),
transforms=transforms)
def __init__(self,
data_shape,
cond_shape,
num_bits,
num_scales,
num_steps,
actnorm,
conditional_channels,
lowres_encoder_channels,
lowres_encoder_blocks,
lowres_encoder_depth,
lowres_upsampler_channels,
pooling,
compression_ratio,
coupling_network,
coupling_blocks,
coupling_channels,
coupling_dropout=0.0,
coupling_gated_conv=None,
coupling_depth=None,
coupling_mixtures=None,
dequant="flow",
dequant_steps=4,
dequant_context=32,
dequant_blocks=2,
augment_steps=4,
augment_context=32,
augment_blocks=2,
augment_size=None,
checkerboard_scales=[],
tuple_flip=True):
if len(compression_ratio) == 1 and num_scales > 1:
compression_ratio = [compression_ratio[0]] * (num_scales - 1)
assert all([
compression_ratio[s] >= 0.0 and compression_ratio[s] < 1.0
for s in range(num_scales - 1)
])
# initialize context. Only upsample context in ContextInit if latent shape doesn't change during the flow.
context_init = ContextInit(num_bits=num_bits,
in_channels=cond_shape[0],
out_channels=lowres_encoder_channels,
mid_channels=lowres_encoder_channels,
num_blocks=lowres_encoder_blocks,
depth=lowres_encoder_depth,
dropout=coupling_dropout)
transforms = []
current_shape = data_shape
if dequant == 'uniform':
transforms.append(UniformDequantization(num_bits=num_bits))
elif dequant == 'flow':
dequantize_flow = DequantizationFlow(
data_shape=data_shape,
num_bits=num_bits,
num_steps=dequant_steps,
coupling_network=coupling_network,
num_context=dequant_context,
num_blocks=dequant_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
dropout=0.0,
gated_conv=False,
num_mixtures=coupling_mixtures,
checkerboard=True,
tuple_flip=tuple_flip)
transforms.append(
VariationalDequantization(encoder=dequantize_flow,
num_bits=num_bits))
# Change range from [0,1]^D to [-0.5, 0.5]^D
transforms.append(ScalarAffineBijection(shift=-0.5))
# Initial squeezing
if current_shape[1] >= 128 and current_shape[2] >= 128:
# H x W -> 64 x 64
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
if current_shape[1] >= 64 and current_shape[2] >= 64:
# H x W -> 32 x 32
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
if 0 not in checkerboard_scales or (current_shape[1] > 32
and current_shape[2] > 32):
# Only go to 16 x 16 if not doing checkerboard splits first
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
# add in augmentation channels if desired
if augment_size is not None and augment_size > 0:
#transforms.append(Augment(StandardUniform((augment_size, current_shape[1], current_shape[2])), x_size=current_shape[0]))
#transforms.append(Augment(StandardNormal((augment_size, current_shape[1], current_shape[2])), x_size=current_shape[0]))
augment_flow = AugmentFlow(data_shape=current_shape,
augment_size=augment_size,
num_steps=augment_steps,
coupling_network=coupling_network,
mid_channels=coupling_channels,
num_context=augment_context,
num_mixtures=coupling_mixtures,
num_blocks=augment_blocks,
dropout=0.0,
checkerboard=True,
tuple_flip=tuple_flip)
transforms.append(
Augment(encoder=augment_flow, x_size=current_shape[0]))
current_shape = (current_shape[0] + augment_size, current_shape[1],
current_shape[2])
for scale in range(num_scales):
# First and Third scales use checkerboard split pattern
checkerboard = scale in checkerboard_scales
context_out_channels = min(current_shape[0], coupling_channels)
context_out_shape = (context_out_channels, current_shape[1],
current_shape[2] //
2) if checkerboard else (context_out_channels,
current_shape[1],
current_shape[2])
# reshape the context to the current size for all flow steps at this scale
context_upsampler_net = UpsamplerNet(
in_channels=lowres_encoder_channels,
out_shape=context_out_shape,
mid_channels=lowres_upsampler_channels)
transforms.append(
ContextUpsampler(context_net=context_upsampler_net,
direction='forward'))
for step in range(num_steps):
flip = (step % 2 == 0) if tuple_flip else False
if len(conditional_channels) == 0:
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
transforms.append(Conv1x1(current_shape[0]))
else:
if actnorm:
transforms.append(
ConditionalActNormBijection2d(
cond_shape=current_shape,
out_channels=current_shape[0],
mid_channels=conditional_channels))
transforms.append(
ConditionalConv1x1(cond_shape=current_shape,
out_channels=current_shape[0],
mid_channels=conditional_channels,
slogdet_cpu=True))
if coupling_network in ["conv", "densenet"]:
transforms.append(
SRCoupling(x_size=context_out_shape,
y_size=current_shape,
mid_channels=coupling_channels,
depth=coupling_depth,
num_blocks=coupling_blocks,
dropout=coupling_dropout,
gated_conv=coupling_gated_conv,
coupling_network=coupling_network,
checkerboard=checkerboard,
flip=flip))
elif coupling_network == "transformer":
transforms.append(
SRMixtureCoupling(x_size=context_out_shape,
y_size=current_shape,
mid_channels=coupling_channels,
dropout=coupling_dropout,
num_blocks=coupling_blocks,
num_mixtures=coupling_mixtures,
checkerboard=checkerboard,
flip=flip))
# Upsample context (for the previous flows, only if moving in the inverse direction)
transforms.append(
ContextUpsampler(context_net=context_upsampler_net,
direction='inverse'))
if scale < num_scales - 1:
if pooling == 'none' or compression_ratio[scale] == 0.0:
# fully bijective flow with multi-scale architecture
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4,
current_shape[1] // 2,
current_shape[2] // 2)
elif pooling == 'slice':
# slice some of the dimensions (channel-wise) out from further flow steps
unsliced_channels = int(
max(
1, 4 * current_shape[0] *
(1.0 - compression_ratio[scale])))
sliced_channels = int(4 * current_shape[0] -
unsliced_channels)
noise_shape = (sliced_channels, current_shape[1] // 2,
current_shape[2] // 2)
transforms.append(Squeeze2d())
transforms.append(
Slice(StandardNormal(noise_shape),
num_keep=unsliced_channels,
dim=1))
current_shape = (unsliced_channels, current_shape[1] // 2,
current_shape[2] // 2)
elif pooling == 'max':
# max pooling to compress dimensions spatially, h//2 and w//2
noise_shape = (current_shape[0] * 3, current_shape[1] // 2,
current_shape[2] // 2)
decoder = StandardHalfNormal(noise_shape)
transforms.append(
SimpleMaxPoolSurjection2d(decoder=decoder))
current_shape = (current_shape[0], current_shape[1] // 2,
current_shape[2] // 2)
elif pooling == "mvae":
# Compressive flow: reduce the dimensionality of data by 2 (channel-wise)
compressed_channels = max(
1,
int(current_shape[0] *
(1.0 - compression_ratio[scale])))
latent_size = compressed_channels * current_shape[
1] * current_shape[2]
vae_channels = [
current_shape[0] * 2, current_shape[0] * 4,
current_shape[0] * 8
]
encoder = ConditionalNormal(ConvEncoderNet(
in_channels=current_shape[0],
out_channels=latent_size,
mid_channels=vae_channels,
max_pool=True,
batch_norm=True),
split_dim=1)
decoder = ConditionalNormal(ConvDecoderNet(
in_channels=latent_size,
out_shape=(current_shape[0] * 2, current_shape[1],
current_shape[2]),
mid_channels=list(reversed(vae_channels)),
batch_norm=True,
in_lambda=lambda x: x.view(x.shape[0], x.shape[1], 1, 1
)),
split_dim=1)
transforms.append(VAE(encoder=encoder, decoder=decoder))
transforms.append(
Reshape(input_shape=(latent_size, ),
output_shape=(compressed_channels,
current_shape[1],
current_shape[2])))
# after reducing channels with mvae, squeeze to reshape latent space before another sequence of flows
transforms.append(Squeeze2d())
current_shape = (
compressed_channels * 4, # current_shape[0] * 4
current_shape[1] // 2,
current_shape[2] // 2)
else:
raise ValueError(
"pooling argument must be either mvae, slice, max, or none"
)
else:
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
# for reference save the shape output by the bijective flow
self.latent_size = current_shape[0] * current_shape[1] * current_shape[
2]
self.flow_shape = current_shape
super(SRPoolFlow, self).__init__(base_dist=ConvNormal2d(current_shape),
transforms=transforms,
context_init=context_init)
def __init__(self,
data_shape,
num_bits,
num_steps,
coupling_network,
num_context,
num_blocks,
mid_channels,
depth,
growth=None,
dropout=None,
gated_conv=None,
num_mixtures=None):
#context_network_type = "conv"
context_network_type = coupling_network
if context_network_type == "densenet":
context_net = nn.Sequential(
LambdaLayer(lambda x: 2 * x.float() / (2**num_bits - 1) - 1),
DenseBlock(in_channels=data_shape[0],
out_channels=mid_channels,
depth=depth,
growth=growth,
dropout=dropout,
gated_conv=gated_conv,
zero_init=False),
nn.Conv2d(mid_channels,
mid_channels,
kernel_size=2,
stride=2,
padding=0),
DenseBlock(in_channels=mid_channels,
out_channels=num_context,
depth=depth,
growth=growth,
dropout=dropout,
gated_conv=gated_conv,
zero_init=False))
elif context_network_type == "transformer":
layers = [
LambdaLayer(lambda x: 2 * x.float() / (2**num_bits - 1) - 1),
Conv2d(in_channels=data_shape[0],
out_channels=mid_channels,
kernel_size=3,
padding=1),
nn.Conv2d(mid_channels,
mid_channels,
kernel_size=2,
stride=2,
padding=0)
]
for i in range(num_blocks):
layers.append(
ConvAttnBlock(channels=mid_channels,
dropout=0.0,
use_attn=False,
context_channels=None))
layers.append(
Conv2d(in_channels=mid_channels,
out_channels=num_context,
kernel_size=3,
padding=1))
context_net = nn.Sequential(*layers)
elif context_network_type == "conv":
context_net = nn.Sequential(
LambdaLayer(lambda x: 2 * x.float() / (2**num_bits - 1) - 1),
Conv2d(data_shape[0],
mid_channels // 2,
kernel_size=3,
stride=1),
nn.Conv2d(mid_channels // 2,
mid_channels,
kernel_size=2,
stride=2,
padding=0),
GatedConvNet(channels=mid_channels, num_blocks=2, dropout=0.0),
Conv2dZeros(in_channels=mid_channels,
out_channels=num_context))
else:
raise ValueError(
f"Unknown dequantization context_network_type type: {context_network_type}"
)
# layer transformations of the dequantization flow
transforms = []
sample_shape = (data_shape[0] * 4, data_shape[1] // 2,
data_shape[2] // 2)
for i in range(num_steps):
#transforms.append(ActNormBijection2d(sample_shape[0]))
transforms.extend([Conv1x1(sample_shape[0])])
if coupling_network in ["conv", "densenet"]:
transforms.append(
ConditionalCoupling(in_channels=sample_shape[0],
num_context=num_context,
num_blocks=num_blocks,
mid_channels=mid_channels,
depth=depth,
growth=growth,
dropout=dropout,
gated_conv=gated_conv,
coupling_network=coupling_network))
elif coupling_network == "transformer":
transforms.append(
ConditionalMixtureCoupling(in_channels=sample_shape[0],
num_context=num_context,
mid_channels=mid_channels,
num_mixtures=num_mixtures,
num_blocks=num_blocks,
dropout=dropout,
use_attn=False))
else:
raise ValueError(
f"Unknown dequantization coupling network type: {coupling_network}"
)
# Final shuffle of channels, squeeze and sigmoid
transforms.extend([Conv1x1(sample_shape[0]), Unsqueeze2d(), Sigmoid()])
super(DequantizationFlow,
self).__init__(base_dist=ConvNormal2d(sample_shape),
transforms=transforms,
context_init=context_net)
def __init__(self,
data_shape,
num_bits,
base_distributions,
num_scales,
num_steps,
actnorm,
vae_hidden_units,
latent_size,
vae_activation,
coupling_network,
dequant,
dequant_steps,
dequant_context,
coupling_blocks,
coupling_channels,
coupling_dropout,
coupling_growth=None,
coupling_gated_conv=None,
coupling_depth=None,
coupling_mixtures=None):
transforms = []
current_shape = data_shape
if num_steps == 0: num_scales = 0
if dequant == 'uniform' or num_steps == 0 or num_scales == 0:
# no bijective flows defaults to only using uniform dequantization
transforms.append(UniformDequantization(num_bits=num_bits))
elif dequant == 'flow':
dequantize_flow = DequantizationFlow(
data_shape=data_shape,
num_bits=num_bits,
num_steps=dequant_steps,
coupling_network=coupling_network,
num_context=dequant_context,
num_blocks=coupling_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
growth=coupling_growth,
dropout=coupling_dropout,
gated_conv=coupling_gated_conv,
num_mixtures=coupling_mixtures)
transforms.append(
VariationalDequantization(encoder=dequantize_flow,
num_bits=num_bits))
# Change range from [0,1]^D to [-0.5, 0.5]^D
transforms.append(ScalarAffineBijection(shift=-0.5))
# Initial squeeze
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
# Dimension preserving flows
for scale in range(num_scales):
for step in range(num_steps):
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
transforms.append(Conv1x1(current_shape[0]))
if coupling_network in ["conv", "densenet"]:
transforms.append(
Coupling(in_channels=current_shape[0],
num_blocks=coupling_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
growth=coupling_growth,
dropout=coupling_dropout,
gated_conv=coupling_gated_conv,
coupling_network=coupling_network))
else:
transforms.append(
MixtureCoupling(in_channels=current_shape[0],
mid_channels=coupling_channels,
num_mixtures=coupling_mixtures,
num_blocks=coupling_blocks,
dropout=coupling_dropout))
if scale < num_scales - 1:
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
else:
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
# Base distribution for dimension preserving portion of flow
if len(base_distributions) > 1:
if base_distributions[0] == "n":
base0 = StandardNormal(current_shape)
elif base_distributions[0] == "c":
base0 = ConvNormal2d(current_shape)
elif base_distributions[0] == "u":
base0 = StandardUniform(current_shape)
else:
raise ValueError(
"Base distribution must be one of n=Noraml, u=Uniform, or c=ConvNormal"
)
else:
base0 = None
# for reference save the shape output by the bijective flow
self.flow_shape = current_shape
# Non-dimension preserving flows
flat_dim = current_shape[0] * current_shape[1] * current_shape[2]
encoder = ConditionalNormal(
MLP(flat_dim,
2 * latent_size,
hidden_units=vae_hidden_units,
activation=vae_activation,
in_lambda=lambda x: x.view(x.shape[0], flat_dim)))
decoder = ConditionalNormal(MLP(
latent_size,
2 * flat_dim,
hidden_units=list(reversed(vae_hidden_units)),
activation=vae_activation,
out_lambda=lambda x: x.view(x.shape[0], current_shape[0] * 2,
current_shape[1], current_shape[2])),
split_dim=1)
transforms.append(VAE(encoder=encoder, decoder=decoder))
# Base distribution for non-dimension preserving portion of flow
#self.latent_size = latent_size
if base_distributions[-1] == "n":
base1 = StandardNormal((latent_size, ))
elif base_distributions[-1] == "c":
base1 = ConvNormal2d((latent_size, ))
elif base_distributions[-1] == "u":
base1 = StandardUniform((latent_size, ))
else:
raise ValueError(
"Base distribution must be one of n=Noraml, u=Uniform, or c=ConvNormal"
)
super(VAECompressiveFlow, self).__init__(base_dist=[base0, base1],
transforms=transforms)
mid_channels=16,
num_blocks=2,
num_mixtures=k,
dropout=0.2), ElementwiseParams2d(2 + k * 3))
#model = Flow(base_dist=StandardNormal((16,7,7)),
model = Flow(
base_dist=ConvNormal2d((16, 7, 7)),
transforms=[
UniformDequantization(num_bits=8),
#Logit(),
ScalarAffineBijection(shift=-0.5),
Squeeze2d(),
ActNormBijection2d(4),
Conv1x1(4),
LogisticMixtureAffineCouplingBijection(net(4),
num_mixtures=k,
scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4),
Conv1x1(4),
LogisticMixtureAffineCouplingBijection(net(4),
num_mixtures=k,
scale_fn=scale_fn("tanh_exp")),
Squeeze2d(),
ActNormBijection2d(16),
Conv1x1(16),
LogisticMixtureAffineCouplingBijection(net(16),
num_mixtures=k,
scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16),
model = Flow(base_dist=ConvNormal2d((16,7,7)),
transforms=[
UniformDequantization(num_bits=8),
# Augment(StandardUniform((1, 28, 28)), x_size=1),
# AffineCouplingBijection(net(2)), ActNormBijection2d(2), Conv1x1(2),
# AffineCouplingBijection(net(2)), ActNormBijection2d(2), Conv1x1(2),
# Squeeze2d(), Slice(StandardNormal((4, 14, 14)), num_keep=4),
# AffineCouplingBijection(net(4)), ActNormBijection2d(4), Conv1x1(4),
# AffineCouplingBijection(net(4)), ActNormBijection2d(4), Conv1x1(4),
# Squeeze2d(), Slice(StandardNormal((8, 7, 7)), num_keep=8),
# AffineCouplingBijection(net(8)), ActNormBijection2d(8), Conv1x1(8),
# AffineCouplingBijection(net(8)), ActNormBijection2d(8), Conv1x1(8),
#Logit(0.05),
ScalarAffineBijection(shift=-0.5),
Squeeze2d(),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(4), Conv1x1(4), AffineCouplingBijection(net(4), scale_fn=scale_fn("tanh_exp")),
Squeeze2d(),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
ActNormBijection2d(16), Conv1x1(16), AffineCouplingBijection(net(16), scale_fn=scale_fn("tanh_exp")),
def __init__(self, data_shape, num_bits, num_scales, num_steps, actnorm,
pooling, dequant, dequant_steps, dequant_context,
densenet_blocks, densenet_channels, densenet_depth,
densenet_growth, dropout, gated_conv):
transforms = []
current_shape = data_shape
if dequant == 'uniform':
transforms.append(UniformDequantization(num_bits=num_bits))
elif dequant == 'flow':
dequantize_flow = DequantizationFlow(
data_shape=data_shape,
num_bits=num_bits,
num_steps=dequant_steps,
num_context=dequant_context,
num_blocks=densenet_blocks,
mid_channels=densenet_channels,
depth=densenet_depth,
growth=densenet_growth,
dropout=dropout,
gated_conv=gated_conv)
transforms.append(
VariationalDequantization(encoder=dequantize_flow,
num_bits=num_bits))
# Change range from [0,1]^D to [-0.5, 0.5]^D
transforms.append(ScalarAffineBijection(shift=-0.5))
# Initial squeeze
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
# Pooling flows
for scale in range(num_scales):
for step in range(num_steps):
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
transforms.extend([
Conv1x1(current_shape[0]),
Coupling(in_channels=current_shape[0],
num_blocks=densenet_blocks,
mid_channels=densenet_channels,
depth=densenet_depth,
growth=densenet_growth,
dropout=dropout,
gated_conv=gated_conv)
])
if scale < num_scales - 1:
noise_shape = (current_shape[0] * 3, current_shape[1] // 2,
current_shape[2] // 2)
if pooling == 'none':
transforms.append(Squeeze2d())
transforms.append(
Slice(StandardNormal(noise_shape),
num_keep=current_shape[0],
dim=1))
elif pooling == 'max':
decoder = StandardHalfNormal(noise_shape)
transforms.append(
SimpleMaxPoolSurjection2d(decoder=decoder))
current_shape = (current_shape[0], current_shape[1] // 2,
current_shape[2] // 2)
else:
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
super(PoolFlow, self).__init__(base_dist=ConvNormal2d(current_shape),
transforms=transforms)
def perm_norm_bi(channels):
return AffineCouplingBijection(net(channels)), \
ActNormBijection2d(channels), \
Conv1x1(channels)
def __init__(self,
data_shape,
num_bits,
num_scales,
num_steps,
actnorm,
pooling,
compression_ratio,
coupling_network,
coupling_blocks,
coupling_channels,
coupling_dropout=0.0,
coupling_gated_conv=None,
coupling_depth=None,
coupling_mixtures=None,
dequant="flow",
dequant_steps=4,
dequant_context=32,
dequant_blocks=2,
augment_steps=4,
augment_context=32,
augment_blocks=2,
augment_size=None,
checkerboard_scales=[],
tuple_flip=True):
if len(compression_ratio) == 1 and num_scales > 1:
compression_ratio = [compression_ratio[0]] * (num_scales - 1)
assert all([
compression_ratio[s] >= 0 and compression_ratio[s] < 1
for s in range(num_scales - 1)
])
transforms = []
current_shape = data_shape
if dequant == 'uniform':
transforms.append(UniformDequantization(num_bits=num_bits))
elif dequant == 'flow':
dequantize_flow = DequantizationFlow(
data_shape=data_shape,
num_bits=num_bits,
num_steps=dequant_steps,
coupling_network=coupling_network,
num_context=dequant_context,
num_blocks=dequant_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
dropout=0.0,
gated_conv=False,
num_mixtures=coupling_mixtures,
checkerboard=True,
tuple_flip=tuple_flip)
transforms.append(
VariationalDequantization(encoder=dequantize_flow,
num_bits=num_bits))
# Change range from [0,1]^D to [-0.5, 0.5]^D
transforms.append(ScalarAffineBijection(shift=-0.5))
# Initial squeezing
if current_shape[1] >= 128 and current_shape[2] >= 128:
# H x W -> 64 x 64
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
if current_shape[1] >= 64 and current_shape[2] >= 64:
# H x W -> 32 x 32
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
if 0 not in checkerboard_scales or (current_shape[1] > 32
and current_shape[2] > 32):
# Only go to 16 x 16 if not doing checkerboard splits first
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4, current_shape[1] // 2,
current_shape[2] // 2)
# add in augmentation channels if desired
if augment_size is not None and augment_size > 0:
augment_flow = AugmentFlow(data_shape=current_shape,
augment_size=augment_size,
num_steps=augment_steps,
coupling_network=coupling_network,
mid_channels=coupling_channels,
num_context=augment_context,
num_mixtures=coupling_mixtures,
num_blocks=augment_blocks,
dropout=0.0,
checkerboard=True,
tuple_flip=tuple_flip)
transforms.append(
Augment(encoder=augment_flow, x_size=current_shape[0]))
current_shape = (current_shape[0] + augment_size, current_shape[1],
current_shape[2])
for scale in range(num_scales):
checkerboard = scale in checkerboard_scales
for step in range(num_steps):
flip = (step % 2 == 0) if tuple_flip else False
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
transforms.append(Conv1x1(current_shape[0]))
if coupling_network == "conv":
transforms.append(
Coupling(in_channels=current_shape[0],
num_blocks=coupling_blocks,
mid_channels=coupling_channels,
depth=coupling_depth,
dropout=coupling_dropout,
gated_conv=coupling_gated_conv,
coupling_network=coupling_network,
checkerboard=checkerboard,
flip=flip))
else:
transforms.append(
MixtureCoupling(in_channels=current_shape[0],
mid_channels=coupling_channels,
num_mixtures=coupling_mixtures,
num_blocks=coupling_blocks,
dropout=coupling_dropout,
checkerboard=checkerboard,
flip=flip))
if scale < num_scales - 1:
if pooling in ['bijective', 'none'
] or compression_ratio[scale] == 0.0:
transforms.append(Squeeze2d())
current_shape = (current_shape[0] * 4,
current_shape[1] // 2,
current_shape[2] // 2)
elif pooling == 'slice':
# slice some of the dimensions (channel-wise) out from further flow steps
unsliced_channels = int(
max(1, 4 * current_shape[0] *
(1.0 - sliced_ratio[scale])))
sliced_channels = int(4 * current_shape[0] -
unsliced_channels)
noise_shape = (sliced_channels, current_shape[1] // 2,
current_shape[2] // 2)
transforms.append(Squeeze2d())
transforms.append(
Slice(StandardNormal(noise_shape),
num_keep=unsliced_channels,
dim=1))
current_shape = (unsliced_channels, current_shape[1] // 2,
current_shape[2] // 2)
elif pooling == 'max':
noise_shape = (current_shape[0] * 3, current_shape[1] // 2,
current_shape[2] // 2)
decoder = StandardHalfNormal(noise_shape)
transforms.append(
SimpleMaxPoolSurjection2d(decoder=decoder))
current_shape = (current_shape[0], current_shape[1] // 2,
current_shape[2] // 2)
else:
raise ValueError(
f"Pooling argument must be either slice, max or none, not: {pooling}"
)
else:
if actnorm:
transforms.append(ActNormBijection2d(current_shape[0]))
# for reference save the shape output by the bijective flow
self.flow_shape = current_shape
self.latent_size = current_shape[0] * current_shape[1] * current_shape[
2]
super(PoolFlow, self).__init__(base_dist=ConvNormal2d(current_shape),
transforms=transforms)