from abc import abstractmethod import math import numpy as np import torch as th import torch.nn as nn import torch.nn.functional as F from .fp16_util import convert_module_to_f16, convert_module_to_f32 from .basic_ops import ( linear, conv_nd, avg_pool_nd, zero_module, normalization, timestep_embedding, ) from .swin_transformer import BasicLayer try: import xformers import xformers.ops as xop XFORMERS_IS_AVAILBLE = True except: XFORMERS_IS_AVAILBLE = False class TimestepBlock(nn.Module): """ Any module where forward() takes timestep embeddings as a second argument. """ @abstractmethod def forward(self, x, emb): """ Apply the module to `x` given `emb` timestep embeddings. """ class TimestepEmbedSequential(nn.Sequential, TimestepBlock): """ A sequential module that passes timestep embeddings to the children that support it as an extra input. """ def forward(self, x, emb): for layer in self: if isinstance(layer, TimestepBlock): x = layer(x, emb) else: x = layer(x) return x class Upsample(nn.Module): """ An upsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determining if a convolution is applied. :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then upsampling occurs in the inner-two dimensions. """ def __init__(self, channels, use_conv, dims=2, out_channels=None): super().__init__() self.channels = channels self.out_channels = out_channels or channels self.use_conv = use_conv self.dims = dims if use_conv: self.conv = conv_nd(dims, self.channels, self.out_channels, 3, padding=1) def forward(self, x): assert x.shape[1] == self.channels if self.dims == 3: x = F.interpolate( x, (x.shape[2], x.shape[3] * 2, x.shape[4] * 2), mode="nearest" ) else: x = F.interpolate(x, scale_factor=2, mode="nearest") if self.use_conv: x = self.conv(x) return x class Downsample(nn.Module): """ A downsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determining if a convolution is applied. :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then downsampling occurs in the inner-two dimensions. """ def __init__(self, channels, use_conv, dims=2, out_channels=None): super().__init__() self.channels = channels self.out_channels = out_channels or channels self.use_conv = use_conv self.dims = dims stride = 2 if dims != 3 else (1, 2, 2) if use_conv: self.op = conv_nd( dims, self.channels, self.out_channels, 3, stride=stride, padding=1 ) else: assert self.channels == self.out_channels self.op = avg_pool_nd(dims, kernel_size=stride, stride=stride) def forward(self, x): assert x.shape[1] == self.channels return self.op(x) class ResBlock(TimestepBlock): """ A residual block that can optionally change the number of channels. :param channels: the number of input channels. :param emb_channels: the number of timestep embedding channels. :param dropout: the rate of dropout. :param out_channels: if specified, the number of out channels. :param use_conv: if True and out_channels is specified, use a spatial convolution instead of a smaller 1x1 convolution to change the channels in the skip connection. :param dims: determines if the signal is 1D, 2D, or 3D. :param up: if True, use this block for upsampling. :param down: if True, use this block for downsampling. """ def __init__( self, channels, emb_channels, dropout, out_channels=None, use_conv=False, use_scale_shift_norm=False, dims=2, up=False, down=False, ): super().__init__() self.channels = channels self.emb_channels = emb_channels self.dropout = dropout self.out_channels = out_channels or channels self.use_conv = use_conv self.use_scale_shift_norm = use_scale_shift_norm self.in_layers = nn.Sequential( normalization(channels), nn.SiLU(), conv_nd(dims, channels, self.out_channels, 3, padding=1), ) self.updown = up or down if up: self.h_upd = Upsample(channels, False, dims) self.x_upd = Upsample(channels, False, dims) elif down: self.h_upd = Downsample(channels, False, dims) self.x_upd = Downsample(channels, False, dims) else: self.h_upd = self.x_upd = nn.Identity() self.emb_layers = nn.Sequential( nn.SiLU(), linear( emb_channels, 2 * self.out_channels if use_scale_shift_norm else self.out_channels, ), ) self.out_layers = nn.Sequential( normalization(self.out_channels), nn.SiLU(), nn.Dropout(p=dropout), zero_module( conv_nd(dims, self.out_channels, self.out_channels, 3, padding=1) ), ) if self.out_channels == channels: self.skip_connection = nn.Identity() elif use_conv: self.skip_connection = conv_nd( dims, channels, self.out_channels, 3, padding=1 ) else: self.skip_connection = conv_nd(dims, channels, self.out_channels, 1) def forward(self, x, emb): if self.updown: in_rest, in_conv = self.in_layers[:-1], self.in_layers[-1] h = in_rest(x) h = self.h_upd(h) x = self.x_upd(x) h = in_conv(h) else: h = self.in_layers(x) emb_out = self.emb_layers(emb).type(h.dtype) while len(emb_out.shape) < len(h.shape): emb_out = emb_out[..., None] if self.use_scale_shift_norm: out_norm, out_rest = self.out_layers[0], self.out_layers[1:] scale, shift = th.chunk(emb_out, 2, dim=1) h = out_norm(h) * (1 + scale) + shift h = out_rest(h) else: h = h + emb_out h = self.out_layers(h) return self.skip_connection(x) + h def count_flops_attn(model, _x, y): """ A counter for the `thop` package to count the operations in an attention operation. Meant to be used like: macs, params = thop.profile( model, inputs=(inputs, timestamps), custom_ops={QKVAttention: QKVAttention.count_flops}, ) """ b, c, *spatial = y[0].shape num_spatial = int(np.prod(spatial)) matmul_ops = 2 * b * (num_spatial ** 2) * c model.total_ops += th.DoubleTensor([matmul_ops]) class AttentionBlock(nn.Module): """ An attention block that allows spatial positions to attend to each other. Originally ported from here, but adapted to the N-d case. https://github.com/hojonathanho/diffusion/blob/1e0dceb3b3495bbe19116a5e1b3596cd0706c543/diffusion_tf/models/unet.py#L66. """ def __init__( self, channels, num_heads=1, num_head_channels=-1, use_new_attention_order=False, ): super().__init__() self.channels = channels if num_head_channels == -1: self.num_heads = num_heads else: assert ( channels % num_head_channels == 0 ), f"q,k,v channels {channels} is not divisible by num_head_channels {num_head_channels}" self.num_heads = channels // num_head_channels self.norm = normalization(channels) self.qkv = conv_nd(1, channels, channels * 3, 1) if use_new_attention_order: # split qkv before split heads self.attention = QKVAttention(self.num_heads) else: # split heads before split qkv self.attention = QKVAttentionLegacy(self.num_heads) self.proj_out = zero_module(conv_nd(1, channels, channels, 1)) def forward(self, x): b, c, *spatial = x.shape x = x.reshape(b, c, -1) qkv = self.qkv(self.norm(x)) h = self.attention(qkv) h = self.proj_out(h) return (x + h).reshape(b, c, *spatial) class QKVAttentionLegacy(nn.Module): """ A module which performs QKV attention. Matches legacy QKVAttention + input/ouput heads shaping """ def __init__(self, n_heads): super().__init__() self.n_heads = n_heads def forward(self, qkv): """ Apply QKV attention. :param qkv: an [N x (H * 3 * C) x T] tensor of Qs, Ks, and Vs. :return: an [N x (H * C) x T] tensor after attention. """ bs, width, length = qkv.shape assert width % (3 * self.n_heads) == 0 ch = width // (3 * self.n_heads) if XFORMERS_IS_AVAILBLE: # qkv: b x length x heads x 3ch qkv = qkv.reshape(bs, self.n_heads, ch * 3, length).permute(0, 3, 1, 2).to(memory_format=th.contiguous_format) q, k, v = qkv.split(ch, dim=3) # b x length x heads x ch a = xop.memory_efficient_attention(q, k, v, p=0.0) # b x length x heads x ch out = a.permute(0, 2, 3, 1).to(memory_format=th.contiguous_format).reshape(bs, -1, length) else: # q,k, v: (b*heads) x ch x length q, k, v = qkv.reshape(bs * self.n_heads, ch * 3, length).split(ch, dim=1) scale = 1 / math.sqrt(math.sqrt(ch)) weight = th.einsum( "bct,bcs->bts", q * scale, k * scale ) # More stable with f16 than dividing afterwards # (b*heads) x M x M weight = th.softmax(weight.float(), dim=-1).type(weight.dtype) a = th.einsum("bts,bcs->bct", weight, v) # (b*heads) x ch x length out = a.reshape(bs, -1, length) return out @staticmethod def count_flops(model, _x, y): return count_flops_attn(model, _x, y) class QKVAttention(nn.Module): """ A module which performs QKV attention and splits in a different order. """ def __init__(self, n_heads): super().__init__() self.n_heads = n_heads def forward(self, qkv): """ Apply QKV attention. :param qkv: an [N x (3 * H * C) x T] tensor of Qs, Ks, and Vs. :return: an [N x (H * C) x T] tensor after attention. """ bs, width, length = qkv.shape assert width % (3 * self.n_heads) == 0 ch = width // (3 * self.n_heads) if XFORMERS_IS_AVAILBLE: # qkv: b x length x heads x 3ch qkv = qkv.reshape(bs, self.n_heads, ch * 3, length).permute(0, 3, 1, 2).to(memory_format=th.contiguous_format) q, k, v = qkv.split(ch, dim=3) # b x length x heads x ch a = xop.memory_efficient_attention(q, k, v, p=0.0) # b x length x heads x length out = a.permute(0, 2, 3, 1).to(memory_format=th.contiguous_format).reshape(bs, -1, length) else: q, k, v = qkv.chunk(3, dim=1) # b x heads*ch x length scale = 1 / math.sqrt(math.sqrt(ch)) weight = th.einsum( "bct,bcs->bts", (q * scale).view(bs * self.n_heads, ch, length), (k * scale).view(bs * self.n_heads, ch, length), ) # More stable with f16 than dividing afterwards weight = th.softmax(weight.float(), dim=-1).type(weight.dtype) a = th.einsum("bts,bcs->bct", weight, v.reshape(bs * self.n_heads, ch, length)) out = a.reshape(bs, -1, length) return out @staticmethod def count_flops(model, _x, y): return count_flops_attn(model, _x, y) class UNetModel(nn.Module): """ The full UNet model with attention and timestep embedding. :param in_channels: channels in the input Tensor. :param model_channels: base channel count for the model. :param out_channels: channels in the output Tensor. :param num_res_blocks: number of residual blocks per downsample. :param attention_resolutions: a collection of downsample rates at which attention will take place. May be a set, list, or tuple. For example, if this contains 4, then at 4x downsampling, attention will be used. :param dropout: the dropout probability. :param channel_mult: channel multiplier for each level of the UNet. :param conv_resample: if True, use learned convolutions for upsampling and downsampling. :param dims: determines if the signal is 1D, 2D, or 3D. :param num_classes: if specified (as an int), then this model will be class-conditional with `num_classes` classes. :param num_heads: the number of attention heads in each attention layer. :param num_heads_channels: if specified, ignore num_heads and instead use a fixed channel width per attention head. :param use_scale_shift_norm: use a FiLM-like conditioning mechanism. :param resblock_updown: use residual blocks for up/downsampling. :param use_new_attention_order: use a different attention pattern for potentially increased efficiency. """ def __init__( self, image_size, in_channels, model_channels, out_channels, num_res_blocks, attention_resolutions, cond_lq=True, dropout=0, channel_mult=(1, 2, 4, 8), conv_resample=True, dims=2, num_classes=None, use_fp16=False, num_heads=1, num_head_channels=-1, use_scale_shift_norm=False, resblock_updown=False, use_new_attention_order=False, ): super().__init__() if isinstance(num_res_blocks, int): num_res_blocks = [num_res_blocks,] * len(channel_mult) else: assert len(num_res_blocks) == len(channel_mult) self.num_res_blocks = num_res_blocks self.image_size = image_size self.in_channels = in_channels self.model_channels = model_channels self.out_channels = out_channels self.attention_resolutions = attention_resolutions self.dropout = dropout self.channel_mult = channel_mult self.conv_resample = conv_resample self.num_classes = num_classes self.dtype = th.float16 if use_fp16 else th.float32 self.num_heads = num_heads self.num_head_channels = num_head_channels self.cond_lq = cond_lq time_embed_dim = model_channels * 4 self.time_embed = nn.Sequential( linear(model_channels, time_embed_dim), nn.SiLU(), linear(time_embed_dim, time_embed_dim), ) if self.num_classes is not None: self.label_emb = nn.Embedding(num_classes, time_embed_dim) ch = input_ch = int(channel_mult[0] * model_channels) self.input_blocks = nn.ModuleList( [TimestepEmbedSequential(conv_nd(dims, in_channels, ch, 3, padding=1))] ) input_block_chans = [ch] ds = image_size for level, mult in enumerate(channel_mult): for _ in range(num_res_blocks[level]): layers = [ ResBlock( ch, time_embed_dim, dropout, out_channels=int(mult * model_channels), dims=dims, use_scale_shift_norm=use_scale_shift_norm, ) ] ch = int(mult * model_channels) if ds in attention_resolutions: layers.append( AttentionBlock( ch, num_heads=num_heads, num_head_channels=num_head_channels, use_new_attention_order=use_new_attention_order, ) ) self.input_blocks.append(TimestepEmbedSequential(*layers)) input_block_chans.append(ch) if level != len(channel_mult) - 1: out_ch = ch self.input_blocks.append( TimestepEmbedSequential( ResBlock( ch, time_embed_dim, dropout, out_channels=out_ch, dims=dims, use_scale_shift_norm=use_scale_shift_norm, down=True, ) if resblock_updown else Downsample( ch, conv_resample, dims=dims, out_channels=out_ch ) ) ) ch = out_ch input_block_chans.append(ch) ds //= 2 self.middle_block = TimestepEmbedSequential( ResBlock( ch, time_embed_dim, dropout, dims=dims, use_scale_shift_norm=use_scale_shift_norm, ), AttentionBlock( ch, num_heads=num_heads, num_head_channels=num_head_channels, use_new_attention_order=use_new_attention_order, ), ResBlock( ch, time_embed_dim, dropout, dims=dims, use_scale_shift_norm=use_scale_shift_norm, ), ) self.output_blocks = nn.ModuleList([]) for level, mult in list(enumerate(channel_mult))[::-1]: for i in range(num_res_blocks[level] + 1): ich = input_block_chans.pop() layers = [ ResBlock( ch + ich, time_embed_dim, dropout, out_channels=int(model_channels * mult), dims=dims, use_scale_shift_norm=use_scale_shift_norm, ) ] ch = int(model_channels * mult) if ds in attention_resolutions: layers.append( AttentionBlock( ch, num_head_channels=num_head_channels, use_new_attention_order=use_new_attention_order, ) ) if level and i == num_res_blocks[level]: out_ch = ch layers.append( ResBlock( ch, time_embed_dim, dropout, out_channels=out_ch, dims=dims, use_scale_shift_norm=use_scale_shift_norm, up=True, ) if resblock_updown else Upsample(ch, conv_resample, dims=dims, out_channels=out_ch) ) ds *= 2 self.output_blocks.append(TimestepEmbedSequential(*layers)) self.out = nn.Sequential( normalization(ch), nn.SiLU(), conv_nd(dims, input_ch, out_channels, 3, padding=1), ) def forward(self, x, timesteps, y=None, lq=None): """ Apply the model to an input batch. :param x: an [N x C x ...] Tensor of inputs. :param timesteps: a 1-D batch of timesteps. :param y: an [N] Tensor of labels, if class-conditional. :param lq: an [N x C x ...] Tensor of low quality iamge. :return: an [N x C x ...] Tensor of outputs. """ assert (y is not None) == ( self.num_classes is not None ), "must specify y if and only if the model is class-conditional" hs = [] emb = self.time_embed(timestep_embedding(timesteps, self.model_channels)).type(self.dtype) if self.num_classes is not None: assert y.shape == (x.shape[0],) emb = emb + self.label_emb(y) if lq is not None: assert self.cond_lq if lq.shape[2:] != x.shape[2:]: lq = F.pixel_unshuffle(lq, 2) x = th.cat([x, lq], dim=1) h = x.type(self.dtype) for ii, module in enumerate(self.input_blocks): h = module(h, emb) hs.append(h) h = self.middle_block(h, emb) for module in self.output_blocks: h = th.cat([h, hs.pop()], dim=1) h = module(h, emb) h = h.type(x.dtype) out = self.out(h) return out def convert_to_fp16(self): """ Convert the torso of the model to float16. """ self.input_blocks.apply(convert_module_to_f16) self.middle_block.apply(convert_module_to_f16) self.output_blocks.apply(convert_module_to_f16) def convert_to_fp32(self): """ Convert the torso of the model to float32. """ self.input_blocks.apply(convert_module_to_f32) self.middle_block.apply(convert_module_to_f32) self.output_blocks.apply(convert_module_to_f32) class UNetModelSwin(nn.Module): """ The full UNet model with attention and timestep embedding. :param in_channels: channels in the input Tensor. :param model_channels: base channel count for the model. :param out_channels: channels in the output Tensor. :param num_res_blocks: number of residual blocks per downsample. :param attention_resolutions: a collection of downsample rates at which attention will take place. May be a set, list, or tuple. For example, if this contains 4, then at 4x downsampling, attention will be used. :param dropout: the dropout probability. :param channel_mult: channel multiplier for each level of the UNet. :param conv_resample: if True, use learned convolutions for upsampling and downsampling. :param dims: determines if the signal is 1D, 2D, or 3D. :param num_classes: if specified (as an int), then this model will be class-conditional with `num_classes` classes. :param num_heads: the number of attention heads in each attention layer. :param num_heads_channels: if specified, ignore num_heads and instead use a fixed channel width per attention head. :param use_scale_shift_norm: use a FiLM-like conditioning mechanism. :param resblock_updown: use residual blocks for up/downsampling. :param use_new_attention_order: use a different attention pattern for potentially increased efficiency. :patch_norm: patch normalization in swin transformer :swin_embed_norm: embed_dim in swin transformer """ def __init__( self, image_size, in_channels, model_channels, out_channels, num_res_blocks, attention_resolutions, dropout=0, channel_mult=(1, 2, 4, 8), conv_resample=True, dims=2, use_fp16=False, num_heads=1, num_head_channels=-1, use_scale_shift_norm=False, resblock_updown=False, swin_depth=2, swin_embed_dim=96, window_size=8, mlp_ratio=2.0, patch_norm=False, cond_lq=True, cond_mask=False, lq_size=256, ): super().__init__() if isinstance(num_res_blocks, int): num_res_blocks = [num_res_blocks,] * len(channel_mult) else: assert len(num_res_blocks) == len(channel_mult) if num_heads == -1: assert swin_embed_dim % num_head_channels == 0 and num_head_channels > 0 self.num_res_blocks = num_res_blocks self.image_size = image_size self.in_channels = in_channels self.model_channels = model_channels self.out_channels = out_channels self.attention_resolutions = attention_resolutions self.dropout = dropout self.channel_mult = channel_mult self.conv_resample = conv_resample self.dtype = th.float16 if use_fp16 else th.float32 self.num_heads = num_heads self.num_head_channels = num_head_channels self.cond_lq = cond_lq self.cond_mask = cond_mask time_embed_dim = model_channels * 4 self.time_embed = nn.Sequential( linear(model_channels, time_embed_dim), nn.SiLU(), linear(time_embed_dim, time_embed_dim), ) if cond_lq and lq_size == image_size: self.feature_extractor = nn.Identity() base_chn = 4 if cond_mask else 3 else: feature_extractor = [] feature_chn = 4 if cond_mask else 3 base_chn = 16 for ii in range(int(math.log(lq_size / image_size) / math.log(2))): feature_extractor.append(nn.Conv2d(feature_chn, base_chn, 3, 1, 1)) feature_extractor.append(nn.SiLU()) feature_extractor.append(Downsample(base_chn, True, out_channels=base_chn*2)) base_chn *= 2 feature_chn = base_chn self.feature_extractor = nn.Sequential(*feature_extractor) ch = input_ch = int(channel_mult[0] * model_channels) in_channels += base_chn self.input_blocks = nn.ModuleList( [TimestepEmbedSequential(conv_nd(dims, in_channels, ch, 3, padding=1))] ) input_block_chans = [ch] ds = image_size for level, mult in enumerate(channel_mult): for jj in range(num_res_blocks[level]): layers = [ ResBlock( ch, time_embed_dim, dropout, out_channels=int(mult * model_channels), dims=dims, use_scale_shift_norm=use_scale_shift_norm, ) ] ch = int(mult * model_channels) if ds in attention_resolutions and jj==0: layers.append( BasicLayer( in_chans=ch, embed_dim=swin_embed_dim, num_heads=num_heads if num_head_channels == -1 else swin_embed_dim // num_head_channels, window_size=window_size, depth=swin_depth, img_size=ds, patch_size=1, mlp_ratio=mlp_ratio, qkv_bias=True, qk_scale=None, drop=dropout, attn_drop=0., drop_path=0., use_checkpoint=False, norm_layer=normalization, patch_norm=patch_norm, ) ) self.input_blocks.append(TimestepEmbedSequential(*layers)) input_block_chans.append(ch) if level != len(channel_mult) - 1: out_ch = ch self.input_blocks.append( TimestepEmbedSequential( ResBlock( ch, time_embed_dim, dropout, out_channels=out_ch, dims=dims, use_scale_shift_norm=use_scale_shift_norm, down=True, ) if resblock_updown else Downsample( ch, conv_resample, dims=dims, out_channels=out_ch ) ) ) ch = out_ch input_block_chans.append(ch) ds //= 2 self.middle_block = TimestepEmbedSequential( ResBlock( ch, time_embed_dim, dropout, dims=dims, use_scale_shift_norm=use_scale_shift_norm, ), BasicLayer( in_chans=ch, embed_dim=swin_embed_dim, num_heads=num_heads if num_head_channels == -1 else swin_embed_dim // num_head_channels, window_size=window_size, depth=swin_depth, img_size=ds, patch_size=1, mlp_ratio=mlp_ratio, qkv_bias=True, qk_scale=None, drop=dropout, attn_drop=0., drop_path=0., use_checkpoint=False, norm_layer=normalization, patch_norm=patch_norm, ), ResBlock( ch, time_embed_dim, dropout, dims=dims, use_scale_shift_norm=use_scale_shift_norm, ), ) self.output_blocks = nn.ModuleList([]) for level, mult in list(enumerate(channel_mult))[::-1]: for i in range(num_res_blocks[level] + 1): ich = input_block_chans.pop() layers = [ ResBlock( ch + ich, time_embed_dim, dropout, out_channels=int(model_channels * mult), dims=dims, use_scale_shift_norm=use_scale_shift_norm, ) ] ch = int(model_channels * mult) if ds in attention_resolutions and i==0: layers.append( BasicLayer( in_chans=ch, embed_dim=swin_embed_dim, num_heads=num_heads if num_head_channels == -1 else swin_embed_dim // num_head_channels, window_size=window_size, depth=swin_depth, img_size=ds, patch_size=1, mlp_ratio=mlp_ratio, qkv_bias=True, qk_scale=None, drop=dropout, attn_drop=0., drop_path=0., use_checkpoint=False, norm_layer=normalization, patch_norm=patch_norm, ) ) if level and i == num_res_blocks[level]: out_ch = ch layers.append( ResBlock( ch, time_embed_dim, dropout, out_channels=out_ch, dims=dims, use_scale_shift_norm=use_scale_shift_norm, up=True, ) if resblock_updown else Upsample(ch, conv_resample, dims=dims, out_channels=out_ch) ) ds *= 2 self.output_blocks.append(TimestepEmbedSequential(*layers)) self.out = nn.Sequential( normalization(ch), nn.SiLU(), conv_nd(dims, input_ch, out_channels, 3, padding=1), ) def forward(self, x, timesteps, lq=None, mask=None): """ Apply the model to an input batch. :param x: an [N x C x ...] Tensor of inputs. :param timesteps: a 1-D batch of timesteps. :param lq: an [N x C x ...] Tensor of low quality iamge. :return: an [N x C x ...] Tensor of outputs. """ hs = [] emb = self.time_embed(timestep_embedding(timesteps, self.model_channels)).type(self.dtype) if lq is not None: assert self.cond_lq if mask is not None: assert self.cond_mask lq = th.cat([lq, mask], dim=1) lq = self.feature_extractor(lq.type(self.dtype)) x = th.cat([x, lq], dim=1) h = x.type(self.dtype) for ii, module in enumerate(self.input_blocks): h = module(h, emb) hs.append(h) h = self.middle_block(h, emb) for module in self.output_blocks: h = th.cat([h, hs.pop()], dim=1) h = module(h, emb) h = h.type(x.dtype) out = self.out(h) return out def convert_to_fp16(self): """ Convert the torso of the model to float16. """ self.input_blocks.apply(convert_module_to_f16) self.feature_extractor.apply(convert_module_to_f16) self.middle_block.apply(convert_module_to_f16) self.output_blocks.apply(convert_module_to_f16) def convert_to_fp32(self): """ Convert the torso of the model to float32. """ self.input_blocks.apply(convert_module_to_f32) self.middle_block.apply(convert_module_to_f32) self.output_blocks.apply(convert_module_to_f32) class ResBlockConv(TimestepBlock): """ A residual block that can optionally change the number of channels. :param channels: the number of input channels. :param emb_channels: the number of timestep embedding channels. :param out_channels: if specified, the number of out channels. :param use_conv: if True and out_channels is specified, use a spatial convolution instead of a smaller 1x1 convolution to change the channels in the skip connection. :param dims: determines if the signal is 1D, 2D, or 3D. :param up: if True, use this block for upsampling. :param down: if True, use this block for downsampling. """ def __init__( self, channels, emb_channels, out_channels=None, use_conv=False, use_scale_shift_norm=False, dims=2, up=False, down=False, ): super().__init__() self.channels = channels self.emb_channels = emb_channels self.out_channels = out_channels or channels self.use_conv = use_conv self.use_scale_shift_norm = use_scale_shift_norm self.in_layers = nn.Sequential( nn.SiLU(), conv_nd(dims, channels, self.out_channels, 3, padding=1), ) self.updown = up or down if up: self.h_upd = Upsample(channels, False, dims) self.x_upd = Upsample(channels, False, dims) elif down: self.h_upd = Downsample(channels, False, dims) self.x_upd = Downsample(channels, False, dims) else: self.h_upd = self.x_upd = nn.Identity() self.emb_layers = nn.Sequential( nn.SiLU(), linear( emb_channels, 2 * self.out_channels if use_scale_shift_norm else self.out_channels, ), ) self.out_layers = nn.Sequential( nn.SiLU(), zero_module( conv_nd(dims, self.out_channels, self.out_channels, 3, padding=1) ), ) if self.out_channels == channels: self.skip_connection = nn.Identity() elif use_conv: self.skip_connection = conv_nd( dims, channels, self.out_channels, 3, padding=1 ) else: self.skip_connection = conv_nd(dims, channels, self.out_channels, 1) def forward(self, x, emb): if self.updown: in_rest, in_conv = self.in_layers[:-1], self.in_layers[-1] h = in_rest(x) h = self.h_upd(h) x = self.x_upd(x) h = in_conv(h) else: h = self.in_layers(x) emb_out = self.emb_layers(emb).type(h.dtype) while len(emb_out.shape) < len(h.shape): emb_out = emb_out[..., None] if self.use_scale_shift_norm: out_norm, out_rest = self.out_layers[0], self.out_layers[1:] scale, shift = th.chunk(emb_out, 2, dim=1) h = out_norm(h) * (1 + scale) + shift h = out_rest(h) else: h = h + emb_out h = self.out_layers(h) return self.skip_connection(x) + h class UNetModelConv(nn.Module): """ The full UNet model with attention and timestep embedding. :param in_channels: channels in the input Tensor. :param model_channels: base channel count for the model. :param out_channels: channels in the output Tensor. :param num_res_blocks: number of residual blocks per downsample. :param attention_resolutions: a collection of downsample rates at which attention will take place. May be a set, list, or tuple. For example, if this contains 4, then at 4x downsampling, attention will be used. :param dropout: the dropout probability. :param channel_mult: channel multiplier for each level of the UNet. :param conv_resample: if True, use learned convolutions for upsampling and downsampling. :param dims: determines if the signal is 1D, 2D, or 3D. :param use_scale_shift_norm: use a FiLM-like conditioning mechanism. :param resblock_updown: use residual blocks for up/downsampling. """ def __init__( self, in_channels, model_channels, out_channels, num_res_blocks, cond_lq=True, channel_mult=(1, 2, 4, 8), conv_resample=True, dims=2, use_scale_shift_norm=False, resblock_updown=False, use_fp16=False, ): super().__init__() if isinstance(num_res_blocks, int): num_res_blocks = [num_res_blocks,] * len(channel_mult) else: assert len(num_res_blocks) == len(channel_mult) self.num_res_blocks = num_res_blocks self.dtype = th.float16 if use_fp16 else th.float32 self.in_channels = in_channels self.model_channels = model_channels self.out_channels = out_channels self.channel_mult = channel_mult self.conv_resample = conv_resample self.cond_lq = cond_lq time_embed_dim = model_channels * 4 self.time_embed = nn.Sequential( linear(model_channels, time_embed_dim), nn.SiLU(), linear(time_embed_dim, time_embed_dim), ) ch = input_ch = int(channel_mult[0] * model_channels) self.input_blocks = nn.ModuleList( [TimestepEmbedSequential(conv_nd(dims, in_channels, ch, 3, padding=1))] ) input_block_chans = [ch] for level, mult in enumerate(channel_mult): for _ in range(num_res_blocks[level]): layers = [ ResBlockConv( ch, time_embed_dim, out_channels=int(mult * model_channels), dims=dims, use_scale_shift_norm=use_scale_shift_norm, ) ] ch = int(mult * model_channels) self.input_blocks.append(TimestepEmbedSequential(*layers)) input_block_chans.append(ch) if level != len(channel_mult) - 1: out_ch = ch self.input_blocks.append( TimestepEmbedSequential( ResBlockConv( ch, time_embed_dim, out_channels=out_ch, dims=dims, use_scale_shift_norm=use_scale_shift_norm, down=True, ) if resblock_updown else Downsample( ch, conv_resample, dims=dims, out_channels=out_ch ) ) ) ch = out_ch input_block_chans.append(ch) self.middle_block = TimestepEmbedSequential( ResBlockConv( ch, time_embed_dim, dims=dims, use_scale_shift_norm=use_scale_shift_norm, ), ResBlockConv( ch, time_embed_dim, dims=dims, use_scale_shift_norm=use_scale_shift_norm, ), ) self.output_blocks = nn.ModuleList([]) for level, mult in list(enumerate(channel_mult))[::-1]: for i in range(num_res_blocks[level] + 1): ich = input_block_chans.pop() layers = [ ResBlockConv( ch + ich, time_embed_dim, out_channels=int(model_channels * mult), dims=dims, use_scale_shift_norm=use_scale_shift_norm, ) ] ch = int(model_channels * mult) if level and i == num_res_blocks[level]: out_ch = ch layers.append( ResBlockConv( ch, time_embed_dim, out_channels=out_ch, dims=dims, use_scale_shift_norm=use_scale_shift_norm, up=True, ) if resblock_updown else Upsample(ch, conv_resample, dims=dims, out_channels=out_ch) ) self.output_blocks.append(TimestepEmbedSequential(*layers)) self.out = nn.Sequential( nn.SiLU(), conv_nd(dims, input_ch, out_channels, 3, padding=1), ) def forward(self, x, timesteps, lq=None): """ Apply the model to an input batch. :param x: an [N x C x ...] Tensor of inputs. :param timesteps: a 1-D batch of timesteps. :param lq: an [N x C x ...] Tensor of low quality iamge. :return: an [N x C x ...] Tensor of outputs. """ hs = [] emb = self.time_embed(timestep_embedding(timesteps, self.model_channels)) if lq is not None: assert self.cond_lq if lq.shape[2:] != x.shape[2:]: lq = F.pixel_unshuffle(lq, 2) x = th.cat([x, lq], dim=1) h = x.type(self.dtype) for ii, module in enumerate(self.input_blocks): h = module(h, emb) hs.append(h) h = self.middle_block(h, emb) for module in self.output_blocks: h = th.cat([h, hs.pop()], dim=1) h = module(h, emb) h = h.type(x.dtype) out = self.out(h) return out