torch实现Gated PixelCNN

PixelCNN

import torch
import torch.nn as nn
import torch.nn.functional as F# Pixel CNNclass MaskConv2d(nn.Module):def __init__(self, conv_type, *args, **kwags):super().__init__()assert conv_type in ('A', 'B')self.conv = nn.Conv2d(*args, **kwags)H, W = self.conv.weight.shape[-2:]mask = torch.zeros((H, W), dtype=torch.float32)mask[0:H // 2, :] = 1mask[H // 2, 0:W // 2] = 1if conv_type == 'B':mask[H // 2, W // 2] = 1mask = mask.reshape((1, 1, H, W))self.register_buffer('mask', mask, False)def forward(self, x):self.conv.weight.data *= self.maskconv_res = self.conv(x)return conv_resclass ResidualBlock(nn.Module):def __init__(self, h, bn=True):super().__init__()self.relu = nn.ReLU()self.conv1 = nn.Conv2d(2 * h, h, 1)self.bn1 = nn.BatchNorm2d(h) if bn else nn.Identity()self.conv2 = MaskConv2d('B', h, h, 3, 1, 1)self.bn2 = nn.BatchNorm2d(h) if bn else nn.Identity()self.conv3 = nn.Conv2d(h, 2 * h, 1)self.bn3 = nn.BatchNorm2d(2 * h) if bn else nn.Identity()def forward(self, x):y = self.relu(x)y = self.conv1(y)y = self.bn1(y)y = self.relu(y)y = self.conv2(y)y = self.bn2(y)y = self.relu(y)y = self.conv3(y)y = self.bn3(y)y = y + xreturn yclass PixelCNN(nn.Module):def __init__(self, n_blocks, h, linear_dim, bn=True, color_level=256):super().__init__()self.conv1 = MaskConv2d('A', 1, 2 * h, 7, 1, 3)self.bn1 = nn.BatchNorm2d(2 * h) if bn else nn.Identity()self.residual_blocks = nn.ModuleList()for _ in range(n_blocks):self.residual_blocks.append(ResidualBlock(h, bn))self.relu = nn.ReLU()self.linear1 = nn.Conv2d(2 * h, linear_dim, 1)self.linear2 = nn.Conv2d(linear_dim, linear_dim, 1)self.out = nn.Conv2d(linear_dim, color_level, 1)def forward(self, x):x = self.conv1(x)x = self.bn1(x)for block in self.residual_blocks:x = block(x)x = self.relu(x)x = self.linear1(x)x = self.relu(x)x = self.linear2(x)x = self.out(x)return x

Gated PixelCNN

class VerticalMaskConv2d(nn.Module):def __init__(self, *args, **kwags):super().__init__()self.conv = nn.Conv2d(*args, **kwags)H, W = self.conv.weight.shape[-2:]mask = torch.zeros((H, W), dtype=torch.float32)mask[0:H // 2 + 1] = 1mask = mask.reshape((1, 1, H, W))self.register_buffer('mask', mask, False)def forward(self, x):self.conv.weight.data *= self.maskconv_res = self.conv(x)return conv_resclass HorizontalMaskConv2d(nn.Module):def __init__(self, conv_type, *args, **kwags):super().__init__()assert conv_type in ('A', 'B')self.conv = nn.Conv2d(*args, **kwags)H, W = self.conv.weight.shape[-2:]mask = torch.zeros((H, W), dtype=torch.float32)mask[H // 2, 0:W // 2] = 1if conv_type == 'B':mask[H // 2, W // 2] = 1mask = mask.reshape((1, 1, H, W))self.register_buffer('mask', mask, False)def forward(self, x):self.conv.weight.data *= self.maskconv_res = self.conv(x)return conv_resclass GatedBlock(nn.Module):def __init__(self, conv_type, in_channels, p, bn=True):super().__init__()self.conv_type = conv_typeself.p = pself.v_conv = VerticalMaskConv2d(in_channels, 2 * p, 3, 1, 1)self.bn1 = nn.BatchNorm2d(2 * p) if bn else nn.Identity()self.v_to_h_conv = nn.Conv2d(2 * p, 2 * p, 1)self.bn2 = nn.BatchNorm2d(2 * p) if bn else nn.Identity()self.h_conv = HorizontalMaskConv2d(conv_type, in_channels, 2 * p, 3, 1,1)self.bn3 = nn.BatchNorm2d(2 * p) if bn else nn.Identity()self.h_output_conv = nn.Conv2d(p, p, 1)self.bn4 = nn.BatchNorm2d(p) if bn else nn.Identity()def forward(self, v_input, h_input):v = self.v_conv(v_input)v = self.bn1(v)v_to_h = v[:, :, 0:-1]v_to_h = F.pad(v_to_h, (0, 0, 1, 0))v_to_h = self.v_to_h_conv(v_to_h)v_to_h = self.bn2(v_to_h)v1, v2 = v[:, :self.p], v[:, self.p:]v1 = torch.tanh(v1)v2 = torch.sigmoid(v2)v = v1 * v2h = self.h_conv(h_input)h = self.bn3(h)h = h + v_to_hh1, h2 = h[:, :self.p], h[:, self.p:]h1 = torch.tanh(h1)h2 = torch.sigmoid(h2)h = h1 * h2h = self.h_output_conv(h)h = self.bn4(h)if self.conv_type == 'B':h = h + h_inputreturn v, hclass GatedPixelCNN(nn.Module):def __init__(self, n_blocks, p, linear_dim, bn=True, color_level=256):super().__init__()self.block1 = GatedBlock('A', 1, p, bn)self.blocks = nn.ModuleList()for _ in range(n_blocks):self.blocks.append(GatedBlock('B', p, p, bn))self.relu = nn.ReLU()self.linear1 = nn.Conv2d(p, linear_dim, 1)self.linear2 = nn.Conv2d(linear_dim, linear_dim, 1)self.out = nn.Conv2d(linear_dim, color_level, 1)def forward(self, x):v, h = self.block1(x, x)for block in self.blocks:v, h = block(v, h)x = self.relu(h)x = self.linear1(x)x = self.relu(x)x = self.linear2(x)x = self.out(x)return x