diff --git a/codes/models/archs/RRDBNet_arch.py b/codes/models/archs/RRDBNet_arch.py index 64873c4a..24de959d 100644 --- a/codes/models/archs/RRDBNet_arch.py +++ b/codes/models/archs/RRDBNet_arch.py @@ -29,6 +29,33 @@ class ResidualDenseBlock_5C(nn.Module): x5 = self.conv5(torch.cat((x, x1, x2, x3, x4), 1)) return x5 * 0.2 + x +# 5-channel residual block that uses attention in the convolutions. +class AttentiveResidualDenseBlock_5C(ResidualDenseBlock_5C): + def __init__(self, nf=64, gc=32, num_convs=8, init_temperature=1): + super(AttentiveResidualDenseBlock_5C, self).__init__() + # gc: growth channel, i.e. intermediate channels + self.conv1 = arch_util.DynamicConv2d(nf, gc, 3, 1, 1, bias=bias, num_convs=num_convs, + initial_temperature=init_temperature) + self.conv2 = arch_util.DynamicConv2d(nf + gc, gc, 3, 1, 1, bias=bias, num_convs=num_convs, + initial_temperature=init_temperature) + self.conv3 = arch_util.DynamicConv2d(nf + 2 * gc, gc, 3, 1, 1, bias=bias, num_convs=num_convs, + initial_temperature=init_temperature) + self.conv4 = arch_util.DynamicConv2d(nf + 3 * gc, gc, 3, 1, 1, bias=bias, num_convs=num_convs, + initial_temperature=init_temperature) + self.conv5 = arch_util.DynamicConv2d(nf + 4 * gc, gc, 3, 1, 1, bias=bias, num_convs=num_convs, + initial_temperature=init_temperature) + + # initialization + arch_util.initialize_weights([self.conv1, self.conv2, self.conv3, self.conv4, self.conv5], + 0.1) + + def set_temperature(self, temp): + self.conv1.set_attention_temperature(temp) + self.conv2.set_attention_temperature(temp) + self.conv3.set_attention_temperature(temp) + self.conv4.set_attention_temperature(temp) + self.conv5.set_attention_temperature(temp) + class RRDB(nn.Module): '''Residual in Residual Dense Block''' @@ -45,16 +72,30 @@ class RRDB(nn.Module): out = self.RDB3(out) return out * 0.2 + x +class AttentiveRRDB(RRDB): + def __init__(self, nf, gc=32, num_convs=8, init_temperature=1): + super(RRDB, self).__init__() + self.RDB1 = AttentiveResidualDenseBlock_5C(nf, gc, num_convs, init_temperature) + self.RDB2 = AttentiveResidualDenseBlock_5C(nf, gc, num_convs, init_temperature) + self.RDB3 = AttentiveResidualDenseBlock_5C(nf, gc, num_convs, init_temperature) + + def set_temperature(self, temp): + self.RDB1.set_temperature(temp) + self.RDB2.set_temperature(temp) + self.RDB3.set_temperature(temp) class RRDBNet(nn.Module): - def __init__(self, in_nc, out_nc, nf, nb, gc=32, scale=2, initial_stride=1): + def __init__(self, in_nc, out_nc, nf, nb, gc=32, scale=2, initial_stride=1, + rrdb_block_f=None): super(RRDBNet, self).__init__() - RRDB_block_f = functools.partial(RRDB, nf=nf, gc=gc) + if rrdb_block_f is None: + rrdb_block_f = functools.partial(RRDB, nf=nf, gc=gc) self.scale = scale self.conv_first = nn.Conv2d(in_nc, nf, 7, initial_stride, padding=3, bias=True) - self.RRDB_trunk = arch_util.make_layer(RRDB_block_f, nb) + self.RRDB_trunk, self.rrdb_layers = arch_util.make_layer(rrdb_block_f, nb, True) self.trunk_conv = nn.Conv2d(nf, nf, 3, 1, 1, bias=True) + #### upsampling self.upconv1 = nn.Conv2d(nf, nf, 5, 1, padding=2, bias=True) self.upconv2 = nn.Conv2d(nf, nf, 5, 1, padding=2, bias=True) @@ -63,6 +104,12 @@ class RRDBNet(nn.Module): self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True) + # Sets the softmax temperature of each RRDB layer. Only works if you are using attentive + # convolutions. + def set_temperature(self, temp): + for layer in self.rrdb_layers: + layer.set_temperature(temp) + def forward(self, x): fea = self.conv_first(x) trunk = self.trunk_conv(self.RRDB_trunk(fea)) diff --git a/codes/models/archs/arch_util.py b/codes/models/archs/arch_util.py index bf634f6b..fa0d92f5 100644 --- a/codes/models/archs/arch_util.py +++ b/codes/models/archs/arch_util.py @@ -28,11 +28,238 @@ def initialize_weights(net_l, scale=1): init.constant_(m.bias.data, 0.0) -def make_layer(block, n_layers): +def make_layer(block, n_layers, return_layers=False): layers = [] for _ in range(n_layers): layers.append(block()) - return nn.Sequential(*layers) + if return_layers: + return nn.Sequential(*layers), layers + else: + return nn.Sequential(*layers) + +class DynamicConv2d(nn.Module): + def __init__(self, nf_in_per_conv, nf_out_per_conv, kernel_size, stride=1, pads=0, has_bias=True, num_convs=8, + att_kernel_size=5, att_pads=2, initial_temperature=1): + super(DynamicConv2d, self).__init__() + + # Requirements: input filter count is even, and there are more filters than there are sequences to attend to. + assert nf_in_per_conv % 2 == 0 + assert nf_in_per_conv / 2 > num_convs + + self.nf = nf_out_per_conv + self.num_convs = num_convs + self.conv_list = nn.ModuleList([nn.Conv2d(nf_in_per_conv, nf_out_per_conv, kernel_size, stride, pads, bias=has_bias) for _ in range(num_convs)]) + self.attention_conv1 = nn.Conv2d(nf_in_per_conv, int(nf_in_per_conv/2), att_kernel_size, stride, att_pads, bias=True) + self.att_bn1 = nn.BatchNorm2d(int(nf_in_per_conv/2)) + self.attention_conv2 = nn.Conv2d(int(nf_in_per_conv/2), num_convs, att_kernel_size, 1, att_pads, bias=True) + self.softmax = nn.Softmax(dim=-1) + self.lrelu = nn.LeakyReLU(negative_slope=0.1, inplace=True) + self.temperature = initial_temperature + + def set_attention_temperature(self, temp): + self.temperature = temp + + def forward(self, x, output_attention_weights=False): + # Build up the individual conv components first. + conv_outputs = [] + for conv in self.conv_list: + conv_outputs.append(conv.forward(x)) + conv_outputs = torch.stack(conv_outputs, dim=0).permute(1, 3, 4, 2, 0) + + # Now calculate the attention across those convs. + conv_attention = self.lrelu(self.att_bn1(self.attention_conv1(x))) + conv_attention = self.attention_conv2(conv_attention).permute(0, 2, 3, 1) + conv_attention = self.softmax(conv_attention / self.temperature) + + # conv_outputs shape: (batch, width, height, filters, sequences) + # conv_attention shape: (batch, width, height, sequences) + # We want to format them so that we can matmul them together to produce: + # desired shape: (batch, width, height, filters) + attention_result = torch.einsum("...ij,...j->...i", [conv_outputs, conv_attention]) + + # Remember to shift the filters back into the expected slot. + if output_attention_weights: + return attention_result.permute(0, 3, 1, 2), conv_attention + else: + return attention_result.permute(0, 3, 1, 2) + +def compute_attention_specificity(att_weights, topk=3): + att = att_weights.detach() + vals, indices = torch.topk(att, topk, dim=-1) + avg = torch.sum(vals, dim=-1) + avg = avg.flatten().mean() + return avg.item(), indices.flatten().detach() + +class DynamicConvTestModule(nn.Module): + def __init__(self): + super(DynamicConvTestModule, self).__init__() + self.init_conv = nn.Conv2d(3, 16, 3, 1, 1, bias=True) + self.conv1 = DynamicConv2d(16, 32, 3, stride=2, pads=1, num_convs=4, initial_temperature=10) + self.bn1 = nn.BatchNorm2d(32) + self.conv2 = DynamicConv2d(32, 64, 3, stride=2, pads=1, att_kernel_size=3, att_pads=1, num_convs=8, initial_temperature=10) + self.bn2 = nn.BatchNorm2d(64) + self.conv3 = DynamicConv2d(64, 128, 3, stride=2, pads=1, att_kernel_size=3, att_pads=1, num_convs=16, initial_temperature=10) + self.bn3 = nn.BatchNorm2d(128) + self.relu = nn.ReLU() + self.dense1 = nn.Linear(128 * 4 * 4, 256) + self.dense2 = nn.Linear(256, 100) + self.softmax = nn.Softmax(-1) + + def set_temp(self, temp): + self.conv1.set_attention_temperature(temp) + self.conv2.set_attention_temperature(temp) + self.conv3.set_attention_temperature(temp) + + def forward(self, x): + x = self.init_conv(x) + x, att1 = self.conv1(x, output_attention_weights=True) + x = self.relu(self.bn1(x)) + x, att2 = self.conv2(x, output_attention_weights=True) + x = self.relu(self.bn2(x)) + x, att3 = self.conv3(x, output_attention_weights=True) + x = self.relu(self.bn3(x)) + atts = [att1, att2, att3] + usage_hists = [] + mean = 0 + for a in atts: + m, u = compute_attention_specificity(a) + mean += m + usage_hists.append(u) + mean /= 3 + + x = x.flatten(1) + x = self.relu(self.dense1(x)) + x = self.dense2(x) + # Compute metrics across attention weights. + + return self.softmax(x), mean, usage_hists + + +class StandardConvTestModule(nn.Module): + def __init__(self): + super(StandardConvTestModule, self).__init__() + self.init_conv = nn.Conv2d(3, 16, 3, 1, 1, bias=True) + self.conv1 = nn.Conv2d(16, 64, 3, stride=2, padding=1) + self.bn1 = nn.BatchNorm2d(64) + self.conv2 = nn.Conv2d(64, 128, 3, stride=2, padding=1) + self.bn2 = nn.BatchNorm2d(128) + self.conv3 = nn.Conv2d(128, 256, 3, stride=2, padding=1) + self.bn3 = nn.BatchNorm2d(256) + self.relu = nn.ReLU() + self.dense1 = nn.Linear(256 * 4 * 4, 256) + self.dense2 = nn.Linear(256, 100) + self.softmax = nn.Softmax(-1) + + def set_temp(self, temp): + pass + + def forward(self, x): + x = self.init_conv(x) + x = self.conv1(x) + x = self.relu(self.bn1(x)) + x = self.conv2(x) + x = self.relu(self.bn2(x)) + x = self.conv3(x) + x = self.relu(self.bn3(x)) + + x = x.flatten(1) + x = self.relu(self.dense1(x)) + x = self.dense2(x) + + return self.softmax(x), 0, [] + +import torch.optim as optim +from torchvision import datasets, models, transforms +import tqdm +from torch.utils.tensorboard import SummaryWriter + +def test_dynamic_conv(): + writer = SummaryWriter() + dataset = datasets.ImageFolder("E:\\data\\cifar-100-python\\images\\train", transforms.Compose([ + transforms.Resize(32, 32), + transforms.RandomHorizontalFlip(), + transforms.ToTensor(), + transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) + ])) + batch_size = 256 + temperature = 30 + loader = torch.utils.data.DataLoader(dataset, batch_size=batch_size, shuffle=True, num_workers=4) + device = torch.device("cuda:0") + net = StandardConvTestModule() + net = net.to(device) + net.set_temp(temperature) + initialize_weights(net) + optimizer = optim.Adam(net.parameters(), lr=1e-3) + + # Load state, where necessary. + ''' + netstate, optimstate = torch.load("test_net.pth") + net.load_state_dict(netstate) + optimizer.load_state_dict(optimstate) + ''' + + criterion = nn.CrossEntropyLoss() + step = 0 + running_corrects = 0 + running_att_mean = 0 + running_att_hist = None + for e in range(300): + tq = tqdm.tqdm(loader) + for batch, labels in tq: + batch = batch.to(device) + labels = labels.to(device) + optimizer.zero_grad() + logits, att_mean, att_usage_hist = net.forward(batch) + running_att_mean += att_mean + if running_att_hist is None: + running_att_hist = att_usage_hist + else: + for i in range(len(att_usage_hist)): + running_att_hist[i] = torch.cat([running_att_hist[i], att_usage_hist[i]]).flatten() + loss = criterion(logits, labels) + loss.backward() + + ''' + if step % 50 == 0: + c1_grad_avg = sum([m.weight.grad.abs().mean().item() for m in net.conv1.conv_list._modules.values()]) / len(net.conv1.conv_list._modules) + c1a_grad_avg = (net.conv1.attention_conv1.weight.grad.abs().mean() + net.conv1.attention_conv2.weight.grad.abs().mean()) / 2 + c2_grad_avg = sum([m.weight.grad.abs().mean().item() for m in net.conv2.conv_list._modules.values()]) / len(net.conv2.conv_list._modules) + c2a_grad_avg = (net.conv2.attention_conv1.weight.grad.abs().mean() + net.conv2.attention_conv2.weight.grad.abs().mean()) / 2 + c3_grad_avg = sum([m.weight.grad.abs().mean().item() for m in net.conv3.conv_list._modules.values()]) / len(net.conv3.conv_list._modules) + c3a_grad_avg = (net.conv3.attention_conv1.weight.grad.abs().mean() + net.conv3.attention_conv2.weight.grad.abs().mean()) / 2 + writer.add_scalar("c1_grad_avg", c1_grad_avg, global_step=step) + writer.add_scalar("c2_grad_avg", c2_grad_avg, global_step=step) + writer.add_scalar("c3_grad_avg", c3_grad_avg, global_step=step) + writer.add_scalar("c1a_grad_avg", c1a_grad_avg, global_step=step) + writer.add_scalar("c2a_grad_avg", c2a_grad_avg, global_step=step) + writer.add_scalar("c3a_grad_avg", c3a_grad_avg, global_step=step) + ''' + + optimizer.step() + _, preds = torch.max(logits, 1) + running_corrects += torch.sum(preds == labels.data) + if step % 50 == 0: + print("Step: %i, Loss: %f, acc: %f, att_mean: %f" % (step, loss.item(), running_corrects / (50.0 * batch_size), + running_att_mean / 50.0)) + writer.add_scalar("Loss", loss.item(), global_step=step) + writer.add_scalar("Accuracy", running_corrects / (50.0 * batch_size), global_step=step) + writer.add_scalar("Att Mean", running_att_mean / 50, global_step=step) + for i in range(len(running_att_hist)): + writer.add_histogram("Att Hist %i" % (i,), running_att_hist[i], global_step=step) + writer.flush() + running_corrects = 0 + running_att_mean = 0 + running_att_hist = None + if step % 1000 == 0: + temperature = max(temperature-1, 1) + net.set_temp(temperature) + print("Temperature drop. Now: %i" % (temperature,)) + step += 1 + torch.save((net.state_dict(), optimizer.state_dict()), "test_net_standard.pth") + +if __name__ == '__main__': + test_dynamic_conv() + class ResidualBlock(nn.Module): '''Residual block with BN