forked from mrq/DL-Art-School
541 lines
29 KiB
Python
541 lines
29 KiB
Python
import torch
|
|
from torch import nn
|
|
from switched_conv import BareConvSwitch, compute_attention_specificity, AttentionNorm
|
|
import torch.nn.functional as F
|
|
import functools
|
|
from collections import OrderedDict
|
|
from models.archs.arch_util import ConvBnLelu, ConvGnSilu, ExpansionBlock, ExpansionBlock2, ConjoinBlock
|
|
from models.archs.RRDBNet_arch import ResidualDenseBlock_5C, RRDB
|
|
from models.archs.spinenet_arch import SpineNet
|
|
from switched_conv_util import save_attention_to_image_rgb
|
|
import os
|
|
|
|
|
|
class MultiConvBlock(nn.Module):
|
|
def __init__(self, filters_in, filters_mid, filters_out, kernel_size, depth, scale_init=1, norm=False, weight_init_factor=1):
|
|
assert depth >= 2
|
|
super(MultiConvBlock, self).__init__()
|
|
self.noise_scale = nn.Parameter(torch.full((1,), fill_value=.01))
|
|
self.bnconvs = nn.ModuleList([ConvBnLelu(filters_in, filters_mid, kernel_size, norm=norm, bias=False, weight_init_factor=weight_init_factor)] +
|
|
[ConvBnLelu(filters_mid, filters_mid, kernel_size, norm=norm, bias=False, weight_init_factor=weight_init_factor) for i in range(depth - 2)] +
|
|
[ConvBnLelu(filters_mid, filters_out, kernel_size, activation=False, norm=False, bias=False, weight_init_factor=weight_init_factor)])
|
|
self.scale = nn.Parameter(torch.full((1,), fill_value=scale_init, dtype=torch.float))
|
|
self.bias = nn.Parameter(torch.zeros(1))
|
|
|
|
def forward(self, x, noise=None):
|
|
if noise is not None:
|
|
noise = noise * self.noise_scale
|
|
x = x + noise
|
|
for m in self.bnconvs:
|
|
x = m.forward(x)
|
|
return x * self.scale + self.bias
|
|
|
|
|
|
# VGG-style layer with Conv(stride2)->BN->Activation->Conv->BN->Activation
|
|
# Doubles the input filter count.
|
|
class HalvingProcessingBlock(nn.Module):
|
|
def __init__(self, filters):
|
|
super(HalvingProcessingBlock, self).__init__()
|
|
self.bnconv1 = ConvGnSilu(filters, filters * 2, stride=2, norm=False, bias=False)
|
|
self.bnconv2 = ConvGnSilu(filters * 2, filters * 2, norm=True, bias=False)
|
|
|
|
def forward(self, x):
|
|
x = self.bnconv1(x)
|
|
return self.bnconv2(x)
|
|
|
|
|
|
# This is a classic u-net architecture with the goal of assigning each individual pixel an individual transform
|
|
# switching set.
|
|
class ConvBasisMultiplexer(nn.Module):
|
|
def __init__(self, input_channels, base_filters, reductions, processing_depth, multiplexer_channels, use_gn=True, use_exp2=False):
|
|
super(ConvBasisMultiplexer, self).__init__()
|
|
self.filter_conv = ConvGnSilu(input_channels, base_filters, bias=True)
|
|
self.reduction_blocks = nn.ModuleList([HalvingProcessingBlock(base_filters * 2 ** i) for i in range(reductions)])
|
|
reduction_filters = base_filters * 2 ** reductions
|
|
self.processing_blocks = nn.Sequential(OrderedDict([('block%i' % (i,), ConvGnSilu(reduction_filters, reduction_filters, bias=False)) for i in range(processing_depth)]))
|
|
if use_exp2:
|
|
self.expansion_blocks = nn.ModuleList([ExpansionBlock2(reduction_filters // (2 ** i)) for i in range(reductions)])
|
|
else:
|
|
self.expansion_blocks = nn.ModuleList([ExpansionBlock(reduction_filters // (2 ** i)) for i in range(reductions)])
|
|
|
|
gap = base_filters - multiplexer_channels
|
|
cbl1_out = ((base_filters - (gap // 2)) // 4) * 4 # Must be multiples of 4 to use with group norm.
|
|
self.cbl1 = ConvGnSilu(base_filters, cbl1_out, norm=use_gn, bias=False, num_groups=4)
|
|
cbl2_out = ((base_filters - (3 * gap // 4)) // 4) * 4
|
|
self.cbl2 = ConvGnSilu(cbl1_out, cbl2_out, norm=use_gn, bias=False, num_groups=4)
|
|
self.cbl3 = ConvGnSilu(cbl2_out, multiplexer_channels, bias=True, norm=False)
|
|
|
|
def forward(self, x):
|
|
x = self.filter_conv(x)
|
|
reduction_identities = []
|
|
for b in self.reduction_blocks:
|
|
reduction_identities.append(x)
|
|
x = b(x)
|
|
x = self.processing_blocks(x)
|
|
for i, b in enumerate(self.expansion_blocks):
|
|
x = b(x, reduction_identities[-i - 1])
|
|
|
|
x = self.cbl1(x)
|
|
x = self.cbl2(x)
|
|
x = self.cbl3(x)
|
|
return x
|
|
|
|
|
|
class CachedBackboneWrapper:
|
|
def __init__(self, backbone: nn.Module):
|
|
self.backbone = backbone
|
|
|
|
def __call__(self, *args):
|
|
self.cache = self.backbone(*args)
|
|
return self.cache
|
|
|
|
def get_forward_result(self):
|
|
return self.cache
|
|
|
|
# torch.gather() which operates across 2d images.
|
|
def gather_2d(input, index):
|
|
b, c, h, w = input.shape
|
|
nodim = input.view(b, c, h * w)
|
|
ind_nd = index[:, 0]*w + index[:, 1]
|
|
ind_nd = ind_nd.unsqueeze(1)
|
|
ind_nd = ind_nd.repeat((1, c))
|
|
ind_nd = ind_nd.unsqueeze(2)
|
|
result = torch.gather(nodim, dim=2, index=ind_nd)
|
|
return result.squeeze()
|
|
|
|
|
|
# Computes a linear latent by performing processing on the reference image and returning the filters of a single point,
|
|
# which should be centered on the image patch being processed.
|
|
#
|
|
# Output is base_filters * 8.
|
|
class ReferenceImageBranch(nn.Module):
|
|
def __init__(self, base_filters=64):
|
|
super(ReferenceImageBranch, self).__init__()
|
|
self.filter_conv = ConvGnSilu(4, base_filters, bias=True)
|
|
self.reduction_blocks = nn.ModuleList([HalvingProcessingBlock(base_filters * 2 ** i) for i in range(3)])
|
|
reduction_filters = base_filters * 2 ** 3
|
|
self.processing_blocks = nn.Sequential(OrderedDict([('block%i' % (i,), ConvGnSilu(reduction_filters, reduction_filters, bias=False)) for i in range(4)]))
|
|
|
|
# center_point is a [b,2] long tensor describing the center point of where the patch was taken from the reference
|
|
# image.
|
|
def forward(self, x, center_point):
|
|
x = self.filter_conv(x)
|
|
reduction_identities = []
|
|
for b in self.reduction_blocks:
|
|
reduction_identities.append(x)
|
|
x = b(x)
|
|
x = self.processing_blocks(x)
|
|
return gather_2d(x, center_point // 8)
|
|
|
|
|
|
# This is similar to ConvBasisMultiplexer, except that it takes a linear reference tensor as a second input to
|
|
# provide better results. It also has fixed parameterization in several places
|
|
class ReferencingConvMultiplexer(nn.Module):
|
|
def __init__(self, input_channels, base_filters, multiplexer_channels, use_gn=True):
|
|
super(ReferencingConvMultiplexer, self).__init__()
|
|
self.filter_conv = ConvGnSilu(input_channels, base_filters, bias=True)
|
|
self.ref_proc = nn.Linear(512, 512)
|
|
self.ref_red = nn.Linear(512, base_filters * 2)
|
|
self.feature_norm = torch.nn.InstanceNorm2d(base_filters)
|
|
self.style_norm = torch.nn.InstanceNorm1d(base_filters)
|
|
|
|
self.reduction_blocks = nn.ModuleList([HalvingProcessingBlock(base_filters * 2 ** i) for i in range(3)])
|
|
reduction_filters = base_filters * 2 ** 3
|
|
self.processing_blocks = nn.Sequential(OrderedDict([('block%i' % (i,), ConvGnSilu(reduction_filters, reduction_filters, bias=False)) for i in range(2)]))
|
|
self.expansion_blocks = nn.ModuleList([ExpansionBlock2(reduction_filters // (2 ** i)) for i in range(3)])
|
|
|
|
gap = base_filters - multiplexer_channels
|
|
cbl1_out = ((base_filters - (gap // 2)) // 4) * 4 # Must be multiples of 4 to use with group norm.
|
|
self.cbl1 = ConvGnSilu(base_filters, cbl1_out, norm=use_gn, bias=False, num_groups=4)
|
|
cbl2_out = ((base_filters - (3 * gap // 4)) // 4) * 4
|
|
self.cbl2 = ConvGnSilu(cbl1_out, cbl2_out, norm=use_gn, bias=False, num_groups=4)
|
|
self.cbl3 = ConvGnSilu(cbl2_out, multiplexer_channels, bias=True, norm=False)
|
|
|
|
def forward(self, x, ref):
|
|
# Start by fusing the reference vector and the input. Follows the ADAIn formula.
|
|
x = self.feature_norm(self.filter_conv(x))
|
|
ref = self.ref_proc(ref)
|
|
ref = self.ref_red(ref)
|
|
b, c = ref.shape
|
|
ref = self.style_norm(ref.view(b, 2, c // 2))
|
|
x = x * ref[:, 0, :].unsqueeze(dim=2).unsqueeze(dim=3).expand(x.shape) + ref[:, 1, :].unsqueeze(dim=2).unsqueeze(dim=3).expand(x.shape)
|
|
|
|
reduction_identities = []
|
|
for b in self.reduction_blocks:
|
|
reduction_identities.append(x)
|
|
x = b(x)
|
|
x = self.processing_blocks(x)
|
|
for i, b in enumerate(self.expansion_blocks):
|
|
x = b(x, reduction_identities[-i - 1])
|
|
|
|
x = self.cbl1(x)
|
|
x = self.cbl2(x)
|
|
x = self.cbl3(x)
|
|
return x
|
|
|
|
|
|
class BackboneMultiplexer(nn.Module):
|
|
def __init__(self, backbone: CachedBackboneWrapper, transform_count):
|
|
super(BackboneMultiplexer, self).__init__()
|
|
self.backbone = backbone
|
|
self.proc = nn.Sequential(ConvGnSilu(256, 256, kernel_size=3, bias=True),
|
|
ConvGnSilu(256, 256, kernel_size=3, bias=False))
|
|
self.up1 = nn.Sequential(ConvGnSilu(256, 128, kernel_size=3, bias=False, norm=False, activation=False),
|
|
ConvGnSilu(128, 128, kernel_size=3, bias=False))
|
|
self.up2 = nn.Sequential(ConvGnSilu(128, 64, kernel_size=3, bias=False, norm=False, activation=False),
|
|
ConvGnSilu(64, 64, kernel_size=3, bias=False))
|
|
self.final = ConvGnSilu(64, transform_count, bias=False, norm=False, activation=False)
|
|
|
|
def forward(self, x):
|
|
spine = self.backbone.get_forward_result()
|
|
feat = self.proc(spine[0])
|
|
feat = self.up1(F.interpolate(feat, scale_factor=2, mode="nearest"))
|
|
feat = self.up2(F.interpolate(feat, scale_factor=2, mode="nearest"))
|
|
return self.final(feat)
|
|
|
|
|
|
class ConfigurableSwitchComputer(nn.Module):
|
|
def __init__(self, base_filters, multiplexer_net, pre_transform_block, transform_block, transform_count, attention_norm,
|
|
init_temp=20, add_scalable_noise_to_transforms=False):
|
|
super(ConfigurableSwitchComputer, self).__init__()
|
|
|
|
tc = transform_count
|
|
self.multiplexer = multiplexer_net(tc)
|
|
|
|
if pre_transform_block:
|
|
self.pre_transform = pre_transform_block()
|
|
else:
|
|
self.pre_transform = None
|
|
self.transforms = nn.ModuleList([transform_block() for _ in range(transform_count)])
|
|
self.add_noise = add_scalable_noise_to_transforms
|
|
self.noise_scale = nn.Parameter(torch.full((1,), float(1e-3)))
|
|
|
|
# And the switch itself, including learned scalars
|
|
self.switch = BareConvSwitch(initial_temperature=init_temp, attention_norm=AttentionNorm(transform_count, accumulator_size=16 * transform_count) if attention_norm else None)
|
|
self.switch_scale = nn.Parameter(torch.full((1,), float(1)))
|
|
self.post_switch_conv = ConvBnLelu(base_filters, base_filters, norm=False, bias=True)
|
|
# The post_switch_conv gets a low scale initially. The network can decide to magnify it (or not)
|
|
# depending on its needs.
|
|
self.psc_scale = nn.Parameter(torch.full((1,), float(.1)))
|
|
|
|
def forward(self, x, output_attention_weights=False, identity=None, att_in=None, fixed_scale=1):
|
|
if att_in is None:
|
|
att_in = x
|
|
|
|
if identity is None:
|
|
identity = x
|
|
|
|
if self.add_noise:
|
|
rand_feature = torch.randn_like(x) * self.noise_scale
|
|
x = x + rand_feature
|
|
|
|
if self.pre_transform:
|
|
x = self.pre_transform(x)
|
|
xformed = [t.forward(x) for t in self.transforms]
|
|
if isinstance(att_in, tuple):
|
|
m = self.multiplexer(*att_in)
|
|
else:
|
|
m = self.multiplexer(att_in)
|
|
|
|
|
|
outputs, attention = self.switch(xformed, m, True)
|
|
outputs = identity + outputs * self.switch_scale * fixed_scale
|
|
outputs = outputs + self.post_switch_conv(outputs) * self.psc_scale * fixed_scale
|
|
if output_attention_weights:
|
|
return outputs, attention
|
|
else:
|
|
return outputs
|
|
|
|
def set_temperature(self, temp):
|
|
self.switch.set_attention_temperature(temp)
|
|
|
|
|
|
class ConfigurableSwitchedResidualGenerator2(nn.Module):
|
|
def __init__(self, switch_depth, switch_filters, switch_reductions, switch_processing_layers, trans_counts, trans_kernel_sizes,
|
|
trans_layers, transformation_filters, attention_norm, initial_temp=20, final_temperature_step=50000, heightened_temp_min=1,
|
|
heightened_final_step=50000, upsample_factor=1,
|
|
add_scalable_noise_to_transforms=False):
|
|
super(ConfigurableSwitchedResidualGenerator2, self).__init__()
|
|
switches = []
|
|
self.initial_conv = ConvBnLelu(3, transformation_filters, norm=False, activation=False, bias=True)
|
|
self.upconv1 = ConvBnLelu(transformation_filters, transformation_filters, norm=False, bias=True)
|
|
self.upconv2 = ConvBnLelu(transformation_filters, transformation_filters, norm=False, bias=True)
|
|
self.hr_conv = ConvBnLelu(transformation_filters, transformation_filters, norm=False, bias=True)
|
|
self.final_conv = ConvBnLelu(transformation_filters, 3, norm=False, activation=False, bias=True)
|
|
for _ in range(switch_depth):
|
|
multiplx_fn = functools.partial(ConvBasisMultiplexer, transformation_filters, switch_filters, switch_reductions, switch_processing_layers, trans_counts)
|
|
pretransform_fn = functools.partial(ConvBnLelu, transformation_filters, transformation_filters, norm=False, bias=False, weight_init_factor=.1)
|
|
transform_fn = functools.partial(MultiConvBlock, transformation_filters, int(transformation_filters * 1.5), transformation_filters, kernel_size=trans_kernel_sizes, depth=trans_layers, weight_init_factor=.1)
|
|
switches.append(ConfigurableSwitchComputer(transformation_filters, multiplx_fn,
|
|
pre_transform_block=pretransform_fn, transform_block=transform_fn,
|
|
attention_norm=attention_norm,
|
|
transform_count=trans_counts, init_temp=initial_temp,
|
|
add_scalable_noise_to_transforms=add_scalable_noise_to_transforms))
|
|
|
|
self.switches = nn.ModuleList(switches)
|
|
self.transformation_counts = trans_counts
|
|
self.init_temperature = initial_temp
|
|
self.final_temperature_step = final_temperature_step
|
|
self.heightened_temp_min = heightened_temp_min
|
|
self.heightened_final_step = heightened_final_step
|
|
self.attentions = None
|
|
self.upsample_factor = upsample_factor
|
|
assert self.upsample_factor == 2 or self.upsample_factor == 4
|
|
|
|
def forward(self, x):
|
|
# This is a common bug when evaluating SRG2 generators. It needs to be configured properly in eval mode. Just fail.
|
|
if not self.train:
|
|
assert self.switches[0].switch.temperature == 1
|
|
|
|
x = self.initial_conv(x)
|
|
|
|
self.attentions = []
|
|
for i, sw in enumerate(self.switches):
|
|
x, att = sw.forward(x, True)
|
|
self.attentions.append(att)
|
|
|
|
x = self.upconv1(F.interpolate(x, scale_factor=2, mode="nearest"))
|
|
if self.upsample_factor > 2:
|
|
x = F.interpolate(x, scale_factor=2, mode="nearest")
|
|
x = self.upconv2(x)
|
|
x = self.final_conv(self.hr_conv(x))
|
|
return x, x
|
|
|
|
def set_temperature(self, temp):
|
|
[sw.set_temperature(temp) for sw in self.switches]
|
|
|
|
def update_for_step(self, step, experiments_path='.'):
|
|
if self.attentions:
|
|
temp = max(1,
|
|
1 + self.init_temperature * (self.final_temperature_step - step) / self.final_temperature_step)
|
|
if temp == 1 and self.heightened_final_step and step > self.final_temperature_step and \
|
|
self.heightened_final_step != 1:
|
|
# Once the temperature passes (1) it enters an inverted curve to match the linear curve from above.
|
|
# without this, the attention specificity "spikes" incredibly fast in the last few iterations.
|
|
h_steps_total = self.heightened_final_step - self.final_temperature_step
|
|
h_steps_current = min(step - self.final_temperature_step, h_steps_total)
|
|
# The "gap" will represent the steps that need to be traveled as a linear function.
|
|
h_gap = 1 / self.heightened_temp_min
|
|
temp = h_gap * h_steps_current / h_steps_total
|
|
# Invert temperature to represent reality on this side of the curve
|
|
temp = 1 / temp
|
|
self.set_temperature(temp)
|
|
if step % 50 == 0:
|
|
output_path = os.path.join(experiments_path, "attention_maps", "a%i")
|
|
prefix = "attention_map_%i_%%i.png" % (step,)
|
|
[save_attention_to_image_rgb(output_path % (i,), self.attentions[i], self.transformation_counts, prefix, step) for i in range(len(self.attentions))]
|
|
|
|
def get_debug_values(self, step):
|
|
temp = self.switches[0].switch.temperature
|
|
mean_hists = [compute_attention_specificity(att, 2) for att in self.attentions]
|
|
means = [i[0] for i in mean_hists]
|
|
hists = [i[1].clone().detach().cpu().flatten() for i in mean_hists]
|
|
val = {"switch_temperature": temp}
|
|
for i in range(len(means)):
|
|
val["switch_%i_specificity" % (i,)] = means[i]
|
|
val["switch_%i_histogram" % (i,)] = hists[i]
|
|
return val
|
|
|
|
|
|
# Equivalent to SRG2 - Uses RDB blocks in between two switches.
|
|
class ConfigurableSwitchedResidualGenerator4(nn.Module):
|
|
def __init__(self, switch_filters, switch_reductions, switch_processing_layers, trans_counts, trans_kernel_sizes,
|
|
trans_layers, transformation_filters, attention_norm, initial_temp=20, final_temperature_step=50000, heightened_temp_min=1,
|
|
heightened_final_step=50000, upsample_factor=1,
|
|
add_scalable_noise_to_transforms=False):
|
|
super(ConfigurableSwitchedResidualGenerator4, self).__init__()
|
|
self.initial_conv = ConvBnLelu(3, transformation_filters, norm=False, activation=False, bias=True)
|
|
self.upconv1 = ConvBnLelu(transformation_filters, transformation_filters, norm=False, bias=True)
|
|
self.upconv2 = ConvBnLelu(transformation_filters, transformation_filters, norm=False, bias=True)
|
|
self.hr_conv = ConvBnLelu(transformation_filters, transformation_filters, norm=False, bias=True)
|
|
|
|
multiplx_fn = functools.partial(ConvBasisMultiplexer, transformation_filters, switch_filters, switch_reductions,
|
|
switch_processing_layers, trans_counts)
|
|
half_multiplx_fn = functools.partial(ConvBasisMultiplexer, transformation_filters, switch_filters, switch_reductions,
|
|
switch_processing_layers, trans_counts // 2)
|
|
transform_fn = functools.partial(MultiConvBlock, transformation_filters, int(transformation_filters * 1.5),
|
|
transformation_filters, kernel_size=trans_kernel_sizes, depth=trans_layers,
|
|
weight_init_factor=.1)
|
|
self.rdb1 = RRDB(transformation_filters)
|
|
self.sw1 = ConfigurableSwitchComputer(transformation_filters, multiplx_fn,
|
|
pre_transform_block=None, transform_block=transform_fn,
|
|
attention_norm=attention_norm,
|
|
transform_count=trans_counts, init_temp=initial_temp,
|
|
add_scalable_noise_to_transforms=add_scalable_noise_to_transforms)
|
|
self.rdb2 = RRDB(transformation_filters)
|
|
self.sw2 = ConfigurableSwitchComputer(transformation_filters, half_multiplx_fn,
|
|
pre_transform_block=None, transform_block=transform_fn,
|
|
attention_norm=attention_norm,
|
|
transform_count=trans_counts // 2, init_temp=initial_temp,
|
|
add_scalable_noise_to_transforms=add_scalable_noise_to_transforms)
|
|
self.rdb3 = RRDB(transformation_filters)
|
|
self.sw3 = ConfigurableSwitchComputer(transformation_filters, multiplx_fn,
|
|
pre_transform_block=None, transform_block=transform_fn,
|
|
attention_norm=attention_norm,
|
|
transform_count=trans_counts, init_temp=initial_temp,
|
|
add_scalable_noise_to_transforms=add_scalable_noise_to_transforms)
|
|
self.rdb4 = RRDB(transformation_filters)
|
|
self.switches = [self.sw1, self.sw2, self.sw3]
|
|
|
|
self.final_conv = ConvBnLelu(transformation_filters, 3, norm=False, activation=False, bias=True)
|
|
self.transformation_counts = trans_counts
|
|
self.init_temperature = initial_temp
|
|
self.final_temperature_step = final_temperature_step
|
|
self.heightened_temp_min = heightened_temp_min
|
|
self.heightened_final_step = heightened_final_step
|
|
self.attentions = None
|
|
self.upsample_factor = upsample_factor
|
|
assert self.upsample_factor == 2 or self.upsample_factor == 4
|
|
|
|
def forward(self, x):
|
|
# This is a common bug when evaluating SRG2 generators. It needs to be configured properly in eval mode. Just fail.
|
|
if not self.train:
|
|
assert self.switches[0].switch.temperature == 1
|
|
|
|
x = self.initial_conv(x)
|
|
|
|
x = self.rdb1(x)
|
|
x, a1 = self.sw1(x, True)
|
|
x = self.rdb2(x)
|
|
x, a2 = self.sw2(x, True)
|
|
x = self.rdb3(x)
|
|
x, a3 = self.sw3(x, True)
|
|
x = self.rdb4(x)
|
|
self.attentions = [a1, a2, a3]
|
|
|
|
x = self.upconv1(F.interpolate(x, scale_factor=2, mode="nearest"))
|
|
if self.upsample_factor > 2:
|
|
x = F.interpolate(x, scale_factor=2, mode="nearest")
|
|
x = self.upconv2(x)
|
|
x = self.final_conv(self.hr_conv(x))
|
|
return x, x
|
|
|
|
def set_temperature(self, temp):
|
|
[sw.set_temperature(temp) for sw in self.switches]
|
|
|
|
def update_for_step(self, step, experiments_path='.'):
|
|
if self.attentions:
|
|
temp = max(1,
|
|
1 + self.init_temperature * (self.final_temperature_step - step) / self.final_temperature_step)
|
|
if temp == 1 and self.heightened_final_step and step > self.final_temperature_step and \
|
|
self.heightened_final_step != 1:
|
|
# Once the temperature passes (1) it enters an inverted curve to match the linear curve from above.
|
|
# without this, the attention specificity "spikes" incredibly fast in the last few iterations.
|
|
h_steps_total = self.heightened_final_step - self.final_temperature_step
|
|
h_steps_current = min(step - self.final_temperature_step, h_steps_total)
|
|
# The "gap" will represent the steps that need to be traveled as a linear function.
|
|
h_gap = 1 / self.heightened_temp_min
|
|
temp = h_gap * h_steps_current / h_steps_total
|
|
# Invert temperature to represent reality on this side of the curve
|
|
temp = 1 / temp
|
|
self.set_temperature(temp)
|
|
if step % 50 == 0:
|
|
output_path = os.path.join(experiments_path, "attention_maps", "a%i")
|
|
prefix = "attention_map_%i_%%i.png" % (step,)
|
|
[save_attention_to_image_rgb(output_path % (i,), self.attentions[i], self.transformation_counts, prefix, step) for i in range(len(self.attentions))]
|
|
|
|
def get_debug_values(self, step):
|
|
temp = self.switches[0].switch.temperature
|
|
mean_hists = [compute_attention_specificity(att, 2) for att in self.attentions]
|
|
means = [i[0] for i in mean_hists]
|
|
hists = [i[1].clone().detach().cpu().flatten() for i in mean_hists]
|
|
val = {"switch_temperature": temp}
|
|
for i in range(len(means)):
|
|
val["switch_%i_specificity" % (i,)] = means[i]
|
|
val["switch_%i_histogram" % (i,)] = hists[i]
|
|
return val
|
|
|
|
class Interpolate(nn.Module):
|
|
def __init__(self, factor):
|
|
super(Interpolate, self).__init__()
|
|
self.factor = factor
|
|
|
|
def forward(self, x):
|
|
return F.interpolate(x, scale_factor=self.factor)
|
|
|
|
|
|
class ConfigurableSwitchedResidualGenerator3(nn.Module):
|
|
def __init__(self, base_filters, trans_count, initial_temp=20, final_temperature_step=50000,
|
|
heightened_temp_min=1,
|
|
heightened_final_step=50000, upsample_factor=4):
|
|
super(ConfigurableSwitchedResidualGenerator3, self).__init__()
|
|
self.initial_conv = ConvBnLelu(3, base_filters, norm=False, activation=False, bias=True)
|
|
self.sw_conv = ConvBnLelu(base_filters, base_filters, activation=False, bias=True)
|
|
self.upconv1 = ConvBnLelu(base_filters, base_filters, norm=False, bias=True)
|
|
self.upconv2 = ConvBnLelu(base_filters, base_filters, norm=False, bias=True)
|
|
self.hr_conv = ConvBnLelu(base_filters, base_filters, norm=False, bias=True)
|
|
self.final_conv = ConvBnLelu(base_filters, 3, norm=False, activation=False, bias=True)
|
|
|
|
self.backbone = SpineNet('49', in_channels=3, use_input_norm=True)
|
|
for p in self.backbone.parameters(recurse=True):
|
|
p.requires_grad = False
|
|
self.backbone_wrapper = CachedBackboneWrapper(self.backbone)
|
|
multiplx_fn = functools.partial(BackboneMultiplexer, self.backbone_wrapper)
|
|
pretransform_fn = functools.partial(nn.Sequential, ConvBnLelu(base_filters, base_filters, kernel_size=3, norm=False, activation=False, bias=False))
|
|
transform_fn = functools.partial(MultiConvBlock, base_filters, int(base_filters * 1.5), base_filters, kernel_size=3, depth=4)
|
|
self.switch = ConfigurableSwitchComputer(base_filters, multiplx_fn, pretransform_fn, transform_fn, trans_count, init_temp=initial_temp,
|
|
add_scalable_noise_to_transforms=True, init_scalar=.1)
|
|
|
|
self.transformation_counts = trans_count
|
|
self.init_temperature = initial_temp
|
|
self.final_temperature_step = final_temperature_step
|
|
self.heightened_temp_min = heightened_temp_min
|
|
self.heightened_final_step = heightened_final_step
|
|
self.attentions = None
|
|
self.upsample_factor = upsample_factor
|
|
self.backbone_forward = None
|
|
|
|
def get_forward_results(self):
|
|
return self.backbone_forward
|
|
|
|
def forward(self, x):
|
|
self.backbone_forward = self.backbone_wrapper(F.interpolate(x, scale_factor=2, mode="nearest"))
|
|
|
|
x = self.initial_conv(x)
|
|
|
|
self.attentions = []
|
|
x, att = self.switch(x, output_attention_weights=True)
|
|
self.attentions.append(att)
|
|
|
|
x = self.upconv1(F.interpolate(x, scale_factor=2, mode="nearest"))
|
|
if self.upsample_factor > 2:
|
|
x = F.interpolate(x, scale_factor=2, mode="nearest")
|
|
x = self.upconv2(x)
|
|
return self.final_conv(self.hr_conv(x)),
|
|
|
|
def set_temperature(self, temp):
|
|
self.switch.set_temperature(temp)
|
|
|
|
def update_for_step(self, step, experiments_path='.'):
|
|
if self.attentions:
|
|
temp = max(1,
|
|
1 + self.init_temperature * (self.final_temperature_step - step) / self.final_temperature_step)
|
|
if temp == 1 and self.heightened_final_step and step > self.final_temperature_step and \
|
|
self.heightened_final_step != 1:
|
|
# Once the temperature passes (1) it enters an inverted curve to match the linear curve from above.
|
|
# without this, the attention specificity "spikes" incredibly fast in the last few iterations.
|
|
h_steps_total = self.heightened_final_step - self.final_temperature_step
|
|
h_steps_current = min(step - self.final_temperature_step, h_steps_total)
|
|
# The "gap" will represent the steps that need to be traveled as a linear function.
|
|
h_gap = 1 / self.heightened_temp_min
|
|
temp = h_gap * h_steps_current / h_steps_total
|
|
# Invert temperature to represent reality on this side of the curve
|
|
temp = 1 / temp
|
|
self.set_temperature(temp)
|
|
if step % 50 == 0:
|
|
output_path = os.path.join(experiments_path, "attention_maps", "a%i")
|
|
prefix = "attention_map_%i_%%i.png" % (step,)
|
|
[save_attention_to_image_rgb(output_path % (i,), self.attentions[i], self.transformation_counts, prefix, step) for i in range(len(self.attentions))]
|
|
|
|
def get_debug_values(self, step):
|
|
temp = self.switch.switch.temperature
|
|
mean_hists = [compute_attention_specificity(att, 2) for att in self.attentions]
|
|
means = [i[0] for i in mean_hists]
|
|
hists = [i[1].clone().detach().cpu().flatten() for i in mean_hists]
|
|
val = {"switch_temperature": temp}
|
|
for i in range(len(means)):
|
|
val["switch_%i_specificity" % (i,)] = means[i]
|
|
val["switch_%i_histogram" % (i,)] = hists[i]
|
|
return val
|
|
|