275 lines
12 KiB
Python
275 lines
12 KiB
Python
import itertools
|
|
from random import randrange
|
|
|
|
import torch
|
|
import torch.nn as nn
|
|
import torch.nn.functional as F
|
|
|
|
from models.arch_util import ResBlock, TimestepEmbedSequential, AttentionBlock, build_local_attention_mask
|
|
from models.diffusion.nn import timestep_embedding, normalization, zero_module, conv_nd, linear
|
|
from models.diffusion.unet_diffusion import TimestepBlock
|
|
from trainer.networks import register_model
|
|
from utils.util import checkpoint
|
|
|
|
|
|
def is_latent(t):
|
|
return t.dtype == torch.float
|
|
|
|
|
|
def is_sequence(t):
|
|
return t.dtype == torch.long
|
|
|
|
|
|
class SubBlock(nn.Module):
|
|
def __init__(self, inp_dim, contraction_dim, blk_dim, heads, dropout):
|
|
super().__init__()
|
|
self.dropout = nn.Dropout(p=dropout)
|
|
self.blk_emb_proj = nn.Conv1d(blk_dim, inp_dim, 1)
|
|
self.attn = AttentionBlock(inp_dim, out_channels=contraction_dim, num_heads=heads)
|
|
self.attnorm = nn.GroupNorm(8, contraction_dim)
|
|
self.ff = nn.Conv1d(inp_dim+contraction_dim, contraction_dim, kernel_size=3, padding=1)
|
|
self.ffnorm = nn.GroupNorm(8, contraction_dim)
|
|
self.mask = build_local_attention_mask(n=4000, l=64, fixed_region=8)
|
|
self.mask_initialized = False
|
|
|
|
def forward(self, x, blk_emb):
|
|
if self.mask is not None and not self.mask_initialized:
|
|
self.mask = self.mask.to(x.device)
|
|
self.mask_initialized = True
|
|
blk_enc = self.blk_emb_proj(blk_emb)
|
|
ah = self.dropout(self.attn(torch.cat([blk_enc, x], dim=-1), mask=self.mask))
|
|
ah = ah[:,:,blk_emb.shape[-1]:] # Strip off the blk_emb and re-align with x.
|
|
ah = F.gelu(self.attnorm(ah))
|
|
h = torch.cat([ah, x], dim=1)
|
|
hf = self.dropout(checkpoint(self.ff, h))
|
|
hf = F.gelu(self.ffnorm(hf))
|
|
h = torch.cat([h, hf], dim=1)
|
|
return h
|
|
|
|
|
|
class ConcatAttentionBlock(TimestepBlock):
|
|
def __init__(self, trunk_dim, contraction_dim, heads, dropout):
|
|
super().__init__()
|
|
self.prenorm = nn.GroupNorm(8, trunk_dim)
|
|
self.block1 = SubBlock(trunk_dim, contraction_dim, trunk_dim, heads, dropout)
|
|
self.block2 = SubBlock(trunk_dim+contraction_dim*2, contraction_dim, trunk_dim, heads, dropout)
|
|
self.out = nn.Conv1d(contraction_dim*4, trunk_dim, kernel_size=1, bias=False)
|
|
self.out.weight.data.zero_()
|
|
|
|
def forward(self, x, blk_emb):
|
|
h = self.prenorm(x)
|
|
h = self.block1(h, blk_emb)
|
|
h = self.block2(h, blk_emb)
|
|
h = self.out(h[:,x.shape[1]:])
|
|
return h + x
|
|
|
|
|
|
class ConditioningEncoder(nn.Module):
|
|
def __init__(self,
|
|
spec_dim,
|
|
embedding_dim,
|
|
num_resolutions,
|
|
attn_blocks=6,
|
|
num_attn_heads=4,
|
|
do_checkpointing=False):
|
|
super().__init__()
|
|
attn = []
|
|
self.init = nn.Conv1d(spec_dim, embedding_dim, kernel_size=5, stride=2)
|
|
self.resolution_embedding = nn.Embedding(num_resolutions, embedding_dim)
|
|
self.resolution_embedding.weight.data.mul(.1) # Reduces the relative influence of this embedding from the start.
|
|
for a in range(attn_blocks):
|
|
attn.append(AttentionBlock(embedding_dim, num_attn_heads, do_checkpoint=do_checkpointing))
|
|
attn.append(ResBlock(embedding_dim, dims=1, checkpointing_enabled=do_checkpointing))
|
|
self.attn = nn.Sequential(*attn)
|
|
self.dim = embedding_dim
|
|
self.do_checkpointing = do_checkpointing
|
|
|
|
def forward(self, x, resolution):
|
|
h = self.init(x) + self.resolution_embedding(resolution).unsqueeze(-1)
|
|
h = self.attn(h)
|
|
return h[:, :, :6]
|
|
|
|
|
|
class TransformerDiffusion(nn.Module):
|
|
"""
|
|
A diffusion model composed entirely of stacks of transformer layers. Why would you do it any other way?
|
|
"""
|
|
def __init__(
|
|
self,
|
|
time_embed_dim=256,
|
|
resolution_steps=8,
|
|
max_window=384,
|
|
model_channels=1024,
|
|
contraction_dim=256,
|
|
num_layers=8,
|
|
in_channels=256,
|
|
input_vec_dim=1024,
|
|
out_channels=512, # mean and variance
|
|
num_heads=4,
|
|
dropout=0,
|
|
use_fp16=False,
|
|
# 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.time_embed_dim = time_embed_dim
|
|
self.out_channels = out_channels
|
|
self.dropout = dropout
|
|
self.unconditioned_percentage = unconditioned_percentage
|
|
self.enable_fp16 = use_fp16
|
|
self.resolution_steps = resolution_steps
|
|
self.max_window = max_window
|
|
self.preprocessed = None
|
|
|
|
self.time_embed = nn.Sequential(
|
|
linear(time_embed_dim, time_embed_dim),
|
|
nn.SiLU(),
|
|
linear(time_embed_dim, model_channels),
|
|
)
|
|
self.resolution_embed = nn.Embedding(resolution_steps, model_channels)
|
|
self.conditioning_encoder = ConditioningEncoder(in_channels, model_channels, resolution_steps, num_attn_heads=model_channels//64)
|
|
self.unconditioned_embedding = nn.Parameter(torch.randn(1,model_channels,6))
|
|
self.unconditioned_prior = nn.Parameter(torch.zeros(1,in_channels,1))
|
|
|
|
self.inp_block = conv_nd(1, in_channels+input_vec_dim, model_channels, 3, 1, 1)
|
|
self.layers = TimestepEmbedSequential(*[ConcatAttentionBlock(model_channels, contraction_dim, num_heads, dropout) for _ in range(num_layers)])
|
|
|
|
self.out = nn.Sequential(
|
|
normalization(model_channels),
|
|
nn.SiLU(),
|
|
zero_module(conv_nd(1, model_channels, out_channels, 3, padding=1)),
|
|
)
|
|
|
|
self.debug_codes = {}
|
|
|
|
def get_grad_norm_parameter_groups(self):
|
|
attn1 = list(itertools.chain.from_iterable([lyr.block1.attn.parameters() for lyr in self.layers]))
|
|
attn2 = list(itertools.chain.from_iterable([lyr.block2.attn.parameters() for lyr in self.layers]))
|
|
ff1 = list(itertools.chain.from_iterable([lyr.block1.ff.parameters() for lyr in self.layers]))
|
|
ff2 = list(itertools.chain.from_iterable([lyr.block2.ff.parameters() for lyr in self.layers]))
|
|
blkout_layers = list(itertools.chain.from_iterable([lyr.out.parameters() for lyr in self.layers]))
|
|
groups = {
|
|
'prenorms': list(itertools.chain.from_iterable([lyr.prenorm.parameters() for lyr in self.layers])),
|
|
'blk1_attention_layers': attn1,
|
|
'blk2_attention_layers': attn2,
|
|
'attention_layers': attn1 + attn2,
|
|
'blk1_ff_layers': ff1,
|
|
'blk2_ff_layers': ff2,
|
|
'ff_layers': ff1 + ff2,
|
|
'block_out_layers': blkout_layers,
|
|
'out': list(self.out.parameters()),
|
|
'x_proj': list(self.inp_block.parameters()),
|
|
'layers': list(self.layers.parameters()),
|
|
'time_embed': list(self.time_embed.parameters()),
|
|
'resolution_embed': list(self.resolution_embed.parameters()),
|
|
}
|
|
return groups
|
|
|
|
def input_to_random_resolution_and_window(self, x, x_prior):
|
|
"""
|
|
This function MUST be applied to the target *before* noising. It returns the reduced, re-scoped target as well
|
|
as caches an internal prior for the rescoped target which will be useud in training.
|
|
Args:
|
|
x: Diffusion target
|
|
x_prior: Prior input, which is generally just {x}
|
|
"""
|
|
assert x.shape == x_prior.shape, f'{x.shape} {x_prior.shape}'
|
|
resolution = randrange(1, self.resolution_steps)
|
|
resolution_scale = 2 ** resolution
|
|
s = F.interpolate(x, scale_factor=1/resolution_scale, mode='linear', align_corners=True)
|
|
s_prior = F.interpolate(x_prior, scale_factor=1/resolution_scale, mode='linear', align_corners=True)
|
|
s_diff = s.shape[-1] - self.max_window
|
|
if s_diff > 1:
|
|
start = randrange(0, s_diff)
|
|
s = s[:,:,start:start+self.max_window]
|
|
s_prior = x_prior[:,:,start:start+self.max_window]
|
|
s_prior = F.interpolate(s_prior, scale_factor=.25, mode='linear', align_corners=True)
|
|
s_prior = F.interpolate(s_prior, size=(s.shape[-1],), mode='linear', align_corners=True)
|
|
self.preprocessed = (s_prior, torch.tensor([resolution] * x.shape[0], dtype=torch.long, device=x.device))
|
|
return s
|
|
|
|
def forward(self, x, timesteps, x_prior=None, resolution=None, conditioning_input=None, conditioning_free=False):
|
|
conditioning_input = x_prior if conditioning_input is None else conditioning_input
|
|
|
|
h = x
|
|
if resolution is None:
|
|
assert self.preprocessed is not None, 'Preprocessing function not called.'
|
|
h = x
|
|
h_sub, resolution = self.preprocessed
|
|
self.preprocessed = None
|
|
else:
|
|
h_sub = F.interpolate(x_prior, scale_factor=4, mode='linear', align_corners=True)
|
|
assert h.shape == h_sub.shape, f'{h.shape} {h_sub.shape}'
|
|
|
|
if conditioning_free:
|
|
code_emb = self.unconditioned_embedding.repeat(x.shape[0], x.shape[-1], 1)
|
|
else:
|
|
MIN_COND_LEN = 200
|
|
MAX_COND_LEN = 1200
|
|
if self.training and conditioning_input.shape[-1] > MAX_COND_LEN:
|
|
clen = randrange(MIN_COND_LEN, MAX_COND_LEN)
|
|
gap = conditioning_input.shape[-1] - clen
|
|
cstart = randrange(0, gap)
|
|
conditioning_input = conditioning_input[:,:,cstart:cstart+clen]
|
|
code_emb = self.conditioning_encoder(conditioning_input, resolution)
|
|
|
|
# Mask out the conditioning input and x_prior inputs for whole batch elements, implementing something similar to classifier-free guidance.
|
|
if self.training and self.unconditioned_percentage > 0:
|
|
unconditioned_batches = torch.rand((h.shape[0], 1, 1),
|
|
device=h.device) < self.unconditioned_percentage
|
|
h_sub = torch.where(unconditioned_batches, self.unconditioned_prior.repeat(h_sub.shape[0], 1, h_sub.shape[-1]), h_sub)
|
|
code_emb = torch.where(unconditioned_batches, self.unconditioned_embedding.repeat(code_emb.shape[0], 1, 1), code_emb)
|
|
|
|
with torch.autocast(x.device.type, enabled=self.enable_fp16):
|
|
time_emb = self.time_embed(timestep_embedding(timesteps, self.time_embed_dim))
|
|
res_emb = self.resolution_embed(resolution)
|
|
blk_emb = torch.cat([time_emb.unsqueeze(-1), res_emb.unsqueeze(-1), code_emb], dim=-1)
|
|
h = torch.cat([h, h_sub], dim=1)
|
|
|
|
h = self.inp_block(h)
|
|
for layer in self.layers:
|
|
h = checkpoint(layer, h, blk_emb)
|
|
|
|
h = h.float()
|
|
out = self.out(h)
|
|
|
|
# Defensively involve probabilistic or possibly unused parameters in loss so we don't get DDP errors.
|
|
unused_params = [self.unconditioned_prior, self.unconditioned_embedding]
|
|
extraneous_addition = 0
|
|
for p in unused_params:
|
|
extraneous_addition = extraneous_addition + p.mean()
|
|
out = out + extraneous_addition * 0
|
|
|
|
return out
|
|
|
|
|
|
@register_model
|
|
def register_transformer_diffusion13(opt_net, opt):
|
|
return TransformerDiffusion(**opt_net['kwargs'])
|
|
|
|
|
|
def test_tfd():
|
|
clip = torch.randn(2,256,10336)
|
|
cond = torch.randn(2,256,10336)
|
|
ts = torch.LongTensor([600, 600])
|
|
model = TransformerDiffusion(in_channels=256, model_channels=1024, contraction_dim=512,
|
|
num_heads=512//64, input_vec_dim=256, num_layers=12, dropout=.1,
|
|
unconditioned_percentage=.6)
|
|
for k in range(100):
|
|
x = model.input_to_random_resolution_and_window(clip, x_prior=clip)
|
|
model(x, ts, clip)
|
|
|
|
|
|
def remove_conditioning(sd_path):
|
|
sd = torch.load(sd_path)
|
|
del sd['unconditioned_embedding']
|
|
torch.save(sd, sd_path.replace('.pth', '') + '_fixed.pth')
|
|
|
|
|
|
if __name__ == '__main__':
|
|
remove_conditioning('X:\\dlas\\experiments\\train_music_diffusion_multilevel_sr_pre\\models\\12500_generator.pth')
|
|
test_tfd()
|