f12f0200d6
parametric efficiency improvements and lets try feeding the timestep into the conditioning encoder
372 lines
14 KiB
Python
372 lines
14 KiB
Python
import itertools
|
|
|
|
import torch
|
|
import torch.nn as nn
|
|
import torch.nn.functional as F
|
|
|
|
from models.diffusion.nn import timestep_embedding, normalization, zero_module, conv_nd, linear
|
|
from models.diffusion.unet_diffusion import TimestepEmbedSequential, \
|
|
Downsample, Upsample, TimestepBlock
|
|
from scripts.audio.gen.use_diffuse_tts import ceil_multiple
|
|
from trainer.networks import register_model
|
|
from utils.util import checkpoint, print_network
|
|
|
|
|
|
def is_sequence(t):
|
|
return t.dtype == torch.long
|
|
|
|
|
|
class ResBlock(TimestepBlock):
|
|
def __init__(
|
|
self,
|
|
channels,
|
|
emb_channels,
|
|
dropout,
|
|
out_channels=None,
|
|
dims=2,
|
|
kernel_size=3,
|
|
efficient_config=False,
|
|
use_scale_shift_norm=False,
|
|
):
|
|
super().__init__()
|
|
self.channels = channels
|
|
self.emb_channels = emb_channels
|
|
self.dropout = dropout
|
|
self.out_channels = out_channels or channels
|
|
self.use_scale_shift_norm = use_scale_shift_norm
|
|
padding = {1: 0, 3: 1, 5: 2}[kernel_size]
|
|
eff_kernel = 1 if efficient_config else 3
|
|
eff_padding = 0 if efficient_config else 1
|
|
|
|
self.in_layers = nn.Sequential(
|
|
normalization(channels),
|
|
nn.SiLU(),
|
|
conv_nd(dims, channels, self.out_channels, eff_kernel, padding=eff_padding),
|
|
)
|
|
|
|
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, kernel_size, padding=padding)
|
|
),
|
|
)
|
|
|
|
if self.out_channels == channels:
|
|
self.skip_connection = nn.Identity()
|
|
else:
|
|
self.skip_connection = conv_nd(dims, channels, self.out_channels, eff_kernel, padding=eff_padding)
|
|
|
|
def forward(self, x, emb):
|
|
"""
|
|
Apply the block to a Tensor, conditioned on a timestep embedding.
|
|
|
|
:param x: an [N x C x ...] Tensor of features.
|
|
:param emb: an [N x emb_channels] Tensor of timestep embeddings.
|
|
:return: an [N x C x ...] Tensor of outputs.
|
|
"""
|
|
return checkpoint(
|
|
self._forward, x, emb
|
|
)
|
|
|
|
def _forward(self, x, emb):
|
|
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 = torch.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 StackedResidualBlock(TimestepBlock):
|
|
def __init__(self, channels, emb_channels, dropout):
|
|
super().__init__()
|
|
|
|
self.emb_layers = nn.Sequential(
|
|
nn.SiLU(),
|
|
linear(
|
|
emb_channels,
|
|
2 * channels,
|
|
),
|
|
)
|
|
|
|
gc = channels // 4
|
|
self.initial_norm = nn.GroupNorm(num_groups=8, num_channels=channels)
|
|
for i in range(5):
|
|
out_channels = channels if i == 4 else gc
|
|
self.add_module(
|
|
f'conv{i + 1}',
|
|
nn.Conv1d(channels + i * gc, out_channels, 3, 1, 1))
|
|
if i != 4:
|
|
self.add_module(f'gn{i+1}', nn.GroupNorm(num_groups=8, num_channels=out_channels))
|
|
self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True)
|
|
zero_module(self.conv5)
|
|
self.drop = nn.Dropout(p=dropout)
|
|
|
|
def forward(self, x, emb):
|
|
return checkpoint(self.forward_, x, emb)
|
|
|
|
def forward_(self, x, emb):
|
|
emb_out = self.emb_layers(emb)
|
|
scale, shift = torch.chunk(emb_out, 2, dim=1)
|
|
x0 = self.initial_norm(x) * (1 + scale.unsqueeze(-1)) + shift.unsqueeze(-1)
|
|
x1 = self.lrelu(self.gn1(self.conv1(x0)))
|
|
x2 = self.lrelu(self.gn2(self.conv2(torch.cat((x, x1), 1))))
|
|
x3 = self.lrelu(self.gn3(self.conv3(torch.cat((x, x1, x2), 1))))
|
|
x4 = self.lrelu(self.gn4(self.conv4(torch.cat((x, x1, x2, x3), 1))))
|
|
x5 = self.conv5(torch.cat((x, x1, x2, x3, x4), 1))
|
|
x5 = self.drop(x5)
|
|
|
|
return x5 + x
|
|
|
|
|
|
class DiffusionWaveformGen(nn.Module):
|
|
"""
|
|
The full UNet model with residual blocks and timestep embedding.
|
|
|
|
Customized to be conditioned on an aligned prior derived from a autoregressive
|
|
GPT-style model.
|
|
|
|
:param in_channels: channels in the input Tensor.
|
|
:param in_latent_channels: channels from the input latent.
|
|
: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 dropout: the dropout probability.
|
|
:param channel_mult: channel multiplier for each level of the UNet.
|
|
:param dims: determines if the signal is 1D, 2D, or 3D.
|
|
"""
|
|
|
|
def __init__(
|
|
self,
|
|
model_channels=512,
|
|
in_channels=64,
|
|
in_mel_channels=256,
|
|
conditioning_dim_factor=2,
|
|
out_channels=128, # mean and variance
|
|
dropout=0,
|
|
channel_mult= (1,1.5,2),
|
|
num_res_blocks=(1,1,0),
|
|
token_conditioning_resolutions=(1,4),
|
|
mid_resnet_depth=10,
|
|
use_fp16=False,
|
|
time_embed_dim_multiplier=1,
|
|
# Parameters for regularization.
|
|
unconditioned_percentage=.1, # This implements a mechanism similar to what is used in classifier-free training.
|
|
):
|
|
super().__init__()
|
|
|
|
self.in_channels = in_channels
|
|
self.model_channels = model_channels
|
|
self.out_channels = out_channels
|
|
self.dropout = dropout
|
|
self.channel_mult = channel_mult
|
|
self.unconditioned_percentage = unconditioned_percentage
|
|
self.enable_fp16 = use_fp16
|
|
self.alignment_size = 2 ** (len(channel_mult)+1)
|
|
self.in_mel_channels = in_mel_channels
|
|
|
|
time_embed_dim = model_channels * time_embed_dim_multiplier
|
|
self.time_embed = nn.Sequential(
|
|
linear(model_channels, time_embed_dim),
|
|
nn.SiLU(),
|
|
linear(time_embed_dim, time_embed_dim),
|
|
)
|
|
|
|
conditioning_dim = model_channels * conditioning_dim_factor
|
|
# Either code_converter or latent_converter is used, depending on what type of conditioning data is fed.
|
|
# This model is meant to be able to be trained on both for efficiency purposes - it is far less computationally
|
|
# complex to generate tokens, while generating latents will normally mean propagating through a deep autoregressive
|
|
# transformer network.
|
|
self.mel_converter = nn.Conv1d(in_mel_channels, conditioning_dim, 3, padding=1)
|
|
self.unconditioned_embedding = nn.Parameter(torch.randn(1,conditioning_dim,1))
|
|
|
|
self.input_blocks = nn.ModuleList(
|
|
[
|
|
TimestepEmbedSequential(
|
|
conv_nd(1, in_channels, model_channels, 3, padding=1)
|
|
)
|
|
]
|
|
)
|
|
token_conditioning_blocks = []
|
|
self._feature_size = model_channels
|
|
input_block_chans = [model_channels]
|
|
ch = model_channels
|
|
ds = 1
|
|
|
|
for level, (mult, num_blocks) in enumerate(zip(channel_mult, num_res_blocks)):
|
|
if ds in token_conditioning_resolutions:
|
|
token_conditioning_block = nn.Conv1d(conditioning_dim, ch, 1)
|
|
token_conditioning_block.weight.data *= .02
|
|
self.input_blocks.append(token_conditioning_block)
|
|
token_conditioning_blocks.append(token_conditioning_block)
|
|
|
|
for _ in range(num_blocks):
|
|
layers = [
|
|
ResBlock(
|
|
ch,
|
|
time_embed_dim,
|
|
dropout,
|
|
out_channels=int(mult * model_channels),
|
|
dims=1,
|
|
kernel_size=3,
|
|
use_scale_shift_norm=True,
|
|
)
|
|
]
|
|
ch = int(mult * model_channels)
|
|
self.input_blocks.append(TimestepEmbedSequential(*layers))
|
|
self._feature_size += ch
|
|
input_block_chans.append(ch)
|
|
if level != len(channel_mult) - 1:
|
|
out_ch = ch
|
|
self.input_blocks.append(
|
|
TimestepEmbedSequential(
|
|
Downsample(
|
|
ch, True, dims=1, out_channels=out_ch, factor=2, ksize=3, pad=1
|
|
)
|
|
)
|
|
)
|
|
ch = out_ch
|
|
input_block_chans.append(ch)
|
|
ds *= 2
|
|
self._feature_size += ch
|
|
|
|
self.middle_block = TimestepEmbedSequential(nn.Conv1d(ch+conditioning_dim, ch, kernel_size=1),
|
|
*[StackedResidualBlock(ch, time_embed_dim, dropout) for _ in range(mid_resnet_depth)])
|
|
self._feature_size += ch
|
|
|
|
self.output_blocks = nn.ModuleList([])
|
|
for level, (mult, num_blocks) in list(enumerate(zip(channel_mult, num_res_blocks)))[::-1]:
|
|
for i in range(num_blocks + 1):
|
|
ich = input_block_chans.pop()
|
|
layers = [
|
|
ResBlock(
|
|
ch + ich,
|
|
time_embed_dim,
|
|
dropout,
|
|
out_channels=int(model_channels * mult),
|
|
dims=1,
|
|
kernel_size=3,
|
|
use_scale_shift_norm=True,
|
|
)
|
|
]
|
|
ch = int(model_channels * mult)
|
|
if level and i == num_blocks:
|
|
out_ch = ch
|
|
layers.append(
|
|
Upsample(ch, True, dims=1, out_channels=out_ch, factor=2)
|
|
)
|
|
ds //= 2
|
|
self.output_blocks.append(TimestepEmbedSequential(*layers))
|
|
self._feature_size += ch
|
|
|
|
self.out = nn.Sequential(
|
|
normalization(ch),
|
|
nn.SiLU(),
|
|
zero_module(conv_nd(1, model_channels, out_channels, 3, padding=1)),
|
|
)
|
|
|
|
def get_grad_norm_parameter_groups(self):
|
|
groups = {
|
|
'input_blocks': list(self.input_blocks.parameters()),
|
|
'output_blocks': list(self.output_blocks.parameters()),
|
|
'middle_rrdb': list(self.middle_block.parameters()),
|
|
}
|
|
return groups
|
|
|
|
def fix_alignment(self, x, aligned_conditioning):
|
|
"""
|
|
The UNet requires that the input <x> is a certain multiple of 2, defined by the UNet depth. Enforce this by
|
|
padding both <x> and <aligned_conditioning> before forward propagation and removing the padding before returning.
|
|
"""
|
|
cm = ceil_multiple(x.shape[-1], self.alignment_size)
|
|
if cm != 0:
|
|
pc = (cm-x.shape[-1])/x.shape[-1]
|
|
x = F.pad(x, (0,cm-x.shape[-1]))
|
|
aligned_conditioning = F.pad(aligned_conditioning, (0,int(pc*aligned_conditioning.shape[-1])))
|
|
return x, aligned_conditioning
|
|
|
|
def forward(self, x, timesteps, codes, conditioning_free=False):
|
|
"""
|
|
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 codes: an aligned latent or sequence of tokens providing useful data about the sample to be produced.
|
|
:param conditioning_free: When set, all conditioning inputs (including tokens and conditioning_input) will not be considered.
|
|
:return: an [N x C x ...] Tensor of outputs.
|
|
"""
|
|
# Fix input size to the proper multiple of 2 so we don't get alignment errors going down and back up the U-net.
|
|
orig_x_shape = x.shape[-1]
|
|
x, codes = self.fix_alignment(x, codes)
|
|
|
|
|
|
hs = []
|
|
time_emb = self.time_embed(timestep_embedding(timesteps, self.model_channels))
|
|
|
|
# Note: this block does not need to repeated on inference, since it is not timestep-dependent.
|
|
if conditioning_free:
|
|
code_emb = self.unconditioned_embedding.repeat(x.shape[0], 1, 1)
|
|
else:
|
|
code_emb = self.mel_converter(codes)
|
|
|
|
time_emb = time_emb.float()
|
|
h = x
|
|
for k, module in enumerate(self.input_blocks):
|
|
if isinstance(module, nn.Conv1d):
|
|
h_tok = F.interpolate(module(code_emb), size=(h.shape[-1]), mode='nearest')
|
|
h = h + h_tok
|
|
else:
|
|
h = module(h, time_emb)
|
|
hs.append(h)
|
|
h = torch.cat([h, F.interpolate(code_emb, size=(h.shape[-1]), mode='nearest')], dim=1)
|
|
h = self.middle_block(h, time_emb)
|
|
for module in self.output_blocks:
|
|
h = torch.cat([h, hs.pop()], dim=1)
|
|
h = module(h, time_emb)
|
|
|
|
out = self.out(h)
|
|
|
|
# Involve probabilistic or possibly unused parameters in loss so we don't get DDP errors.
|
|
extraneous_addition = 0
|
|
params = [self.unconditioned_embedding]
|
|
for p in params:
|
|
extraneous_addition = extraneous_addition + p.mean()
|
|
out = out + extraneous_addition * 0
|
|
|
|
return out[:, :, :orig_x_shape]
|
|
|
|
|
|
@register_model
|
|
def register_unet_diffusion_waveform_gen3(opt_net, opt):
|
|
return DiffusionWaveformGen(**opt_net['kwargs'])
|
|
|
|
|
|
if __name__ == '__main__':
|
|
clip = torch.randn(2, 4, 880)
|
|
aligned_sequence = torch.randn(2,256,220)
|
|
ts = torch.LongTensor([600, 600])
|
|
model = DiffusionWaveformGen(in_channels=4, out_channels=8, model_channels=64, in_mel_channels=256,
|
|
channel_mult=[1,2,4,6,8,16], num_res_blocks=[2,2,2,1,1,0], mid_resnet_depth=24,
|
|
conditioning_dim_factor=8,
|
|
token_conditioning_resolutions=[4,16], dropout=.1, time_embed_dim_multiplier=4)
|
|
# Test with sequence aligned conditioning
|
|
o = model(clip, ts, aligned_sequence)
|
|
print_network(model)
|
|
|