deviandice/DL-Art-School
1
0
Fork 0
forked from mrq/DL-Art-School
Code Issues Pull Requests Packages Projects Releases Wiki Activity

prep flat0 for feeding from autoregressive_latent_converter

Browse Source
This commit is contained in:
James Betker 2022-04-01 15:53:45 -06:00
parent 3e97abc8a9
commit f6a8b0a5ca
1 changed files with 19 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
codes/models/audio/tts/unet_diffusion_tts_flat0.py
View File

@ -118,6 +118,7 @@ class DiffusionTtsFlat(nn.Module):
dropout=0,
use_fp16=False,
num_heads=16,
freeze_everything_except_autoregressive_inputs=False,
# Parameters for regularization.
layer_drop=.1,
unconditioned_percentage=.1, # This implements a mechanism similar to what is used in classifier-free training.
@ -151,7 +152,11 @@ class DiffusionTtsFlat(nn.Module):
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
)
self.code_norm = normalization(model_channels)
self.latent_converter = nn.Conv1d(in_latent_channels, model_channels, 1)
self.autoregressive_latent_converter = nn.Sequential(nn.Conv1d(in_latent_channels, model_channels, 1),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
AttentionBlock(model_channels, num_heads, relative_pos_embeddings=True),
)
self.contextual_embedder = nn.Sequential(nn.Conv1d(in_channels,model_channels,3,padding=1,stride=2),
nn.Conv1d(model_channels, model_channels*2,3,padding=1,stride=2),
AttentionBlock(model_channels*2, num_heads, relative_pos_embeddings=True, do_checkpoint=False),
@ -177,11 +182,19 @@ class DiffusionTtsFlat(nn.Module):
zero_module(conv_nd(1, model_channels, out_channels, 3, padding=1)),
)
if freeze_everything_except_autoregressive_inputs:
for ap in list(self.autoregressive_latent_converter.parameters()):
ap.ALLOWED_IN_FLAT = True
for p in self.parameters():
if not hasattr(p, 'ALLOWED_IN_FLAT'):
p.requires_grad = False
p.DO_NOT_TRAIN = True
def get_grad_norm_parameter_groups(self):
groups = {
'minicoder': list(self.contextual_embedder.parameters()),
'layers': list(self.layers.parameters()),
'code_converters': list(self.code_embedding.parameters()) + list(self.code_converter.parameters()) + list(self.latent_converter.parameters()) + list(self.latent_converter.parameters()),
'code_converters': list(self.code_embedding.parameters()) + list(self.code_converter.parameters()) + list(self.autoregressive_latent_converter.parameters()) + list(self.autoregressive_latent_converter.parameters()),
'timestep_integrator': list(self.conditioning_timestep_integrator.parameters()) + list(self.integrating_conv.parameters()),
'time_embed': list(self.time_embed.parameters()),
}
@ -202,7 +215,7 @@ class DiffusionTtsFlat(nn.Module):
cond_emb = conds.mean(dim=-1)
cond_scale, cond_shift = torch.chunk(cond_emb, 2, dim=1)
if is_latent(aligned_conditioning):
code_emb = self.latent_converter(aligned_conditioning)
code_emb = self.autoregressive_latent_converter(aligned_conditioning)
else:
code_emb = self.code_embedding(aligned_conditioning).permute(0, 2, 1)
code_emb = self.code_converter(code_emb)
@ -245,7 +258,7 @@ class DiffusionTtsFlat(nn.Module):
if conditioning_free:
code_emb = self.unconditioned_embedding.repeat(x.shape[0], 1, x.shape[-1])
unused_params.extend(list(self.code_converter.parameters()) + list(self.code_embedding.parameters()))
unused_params.extend(list(self.latent_converter.parameters()))
unused_params.extend(list(self.autoregressive_latent_converter.parameters()))
else:
if precomputed_aligned_embeddings is not None:
code_emb = precomputed_aligned_embeddings
@ -256,7 +269,7 @@ class DiffusionTtsFlat(nn.Module):
if is_latent(aligned_conditioning):
unused_params.extend(list(self.code_converter.parameters()) + list(self.code_embedding.parameters()))
else:
unused_params.extend(list(self.latent_converter.parameters()))
unused_params.extend(list(self.autoregressive_latent_converter.parameters()))
time_emb = self.time_embed(timestep_embedding(timesteps, self.model_channels))
code_emb = self.conditioning_timestep_integrator(code_emb, time_emb)
@ -297,7 +310,7 @@ if __name__ == '__main__':
aligned_sequence = torch.randint(0,8192,(2,100))
cond = torch.randn(2, 100, 400)
ts = torch.LongTensor([600, 600])
model = DiffusionTtsFlat(512, layer_drop=.3, unconditioned_percentage=.5)
model = DiffusionTtsFlat(512, layer_drop=.3, unconditioned_percentage=.5, freeze_everything_except_autoregressive_inputs=True)
# Test with latent aligned conditioning
#o = model(clip, ts, aligned_latent, cond)
# Test with sequence aligned conditioning