From c14fc003ed370ac482669447b150bd100ff59382 Mon Sep 17 00:00:00 2001 From: James Betker Date: Thu, 17 Mar 2022 17:45:27 -0600 Subject: [PATCH] flat diffusion --- .../audio/tts/unet_diffusion_tts_flat.py | 44 ++++++++----------- codes/trainer/eval/audio_diffusion_fid.py | 15 ++++--- 2 files changed, 28 insertions(+), 31 deletions(-) diff --git a/codes/models/audio/tts/unet_diffusion_tts_flat.py b/codes/models/audio/tts/unet_diffusion_tts_flat.py index a7c96527..c910144d 100644 --- a/codes/models/audio/tts/unet_diffusion_tts_flat.py +++ b/codes/models/audio/tts/unet_diffusion_tts_flat.py @@ -101,21 +101,14 @@ class ResBlock(TimestepBlock): class DiffusionLayer(nn.Module): - def __init__(self, model_channels, aligned_channels, cond_channels, dropout, num_heads): + def __init__(self, model_channels, dropout, num_heads): super().__init__() - self.aligned_mutation = zero_module(conv_nd(1, aligned_channels, model_channels, 1)) - self.cond_mutation = zero_module(conv_nd(1, cond_channels, model_channels, 1)) - self.inp_mutation = conv_nd(1, model_channels, model_channels, 1) self.resblk = ResBlock(model_channels, model_channels, dropout, model_channels, dims=1, use_scale_shift_norm=True) self.attn = AttentionBlock(model_channels, num_heads) - def forward(self, x, aligned, pointwise, time_emb): - a = self.aligned_mutation(aligned) - c = self.cond_mutation(pointwise.unsqueeze(-1)) - f = self.inp_mutation(x) - y = self.resblk(f + c.repeat(1,1,f.shape[-1]) + F.interpolate(a, size=f.shape[-1], mode='nearest'), time_emb) - y = self.attn(y) - return y + def forward(self, x, time_emb): + y = self.resblk(x, time_emb) + return self.attn(y) class DiffusionTtsFlat(nn.Module): @@ -147,8 +140,8 @@ class DiffusionTtsFlat(nn.Module): self.enable_fp16 = use_fp16 self.layer_drop = layer_drop - self.inp_block = conv_nd(1, in_channels, model_channels, 3, 1, 1) - self.position_embed = nn.Embedding(max_positions, model_channels) + self.inp_block = conv_nd(1, in_channels, model_channels//2, 3, 1, 1) + self.position_embed = nn.Embedding(max_positions, model_channels//2) self.time_embed = nn.Embedding(max_timesteps, model_channels) # Either code_converter or latent_converter is used, depending on what type of conditioning data is fed. @@ -171,7 +164,8 @@ class DiffusionTtsFlat(nn.Module): ff_glu=True, rotary_emb_dim=True, ) - )) + ) + ) self.latent_converter = nn.Conv1d(in_latent_channels, model_channels, 1) if in_channels > 60: # It's a spectrogram. self.contextual_embedder = nn.Sequential(nn.Conv1d(in_channels,model_channels,3,padding=1,stride=2), @@ -196,14 +190,14 @@ class DiffusionTtsFlat(nn.Module): self.conditioning_conv = nn.Conv1d(model_channels*2, model_channels, 1) self.unconditioned_embedding = nn.Parameter(torch.randn(1,model_channels,1)) self.conditioning_timestep_integrator = TimestepEmbedSequential( - ResBlock(model_channels, model_channels, dropout, out_channels=model_channels, dims=1, kernel_size=1, use_scale_shift_norm=True), - AttentionBlock(model_channels, num_heads=num_heads), - ResBlock(model_channels, model_channels, dropout, out_channels=model_channels, dims=1, kernel_size=1, use_scale_shift_norm=True), - AttentionBlock(model_channels, num_heads=num_heads), - ResBlock(model_channels, model_channels, dropout, out_channels=model_channels, dims=1, kernel_size=1, use_scale_shift_norm=True), + ResBlock(model_channels, model_channels, dropout, out_channels=model_channels, dims=1, kernel_size=1, use_scale_shift_norm=True), + AttentionBlock(model_channels, num_heads=num_heads), + ResBlock(model_channels, model_channels, dropout, out_channels=model_channels, dims=1, kernel_size=1, use_scale_shift_norm=True), + AttentionBlock(model_channels, num_heads=num_heads), + ResBlock(model_channels, model_channels, dropout, out_channels=model_channels//2, dims=1, kernel_size=1, use_scale_shift_norm=True), ) - self.layers = nn.ModuleList([DiffusionLayer(model_channels, model_channels, model_channels, dropout, num_heads) for _ in range(num_layers)]) + self.layers = nn.ModuleList([DiffusionLayer(model_channels, dropout, num_heads) for _ in range(num_layers)]) self.out = nn.Sequential( normalization(model_channels), @@ -259,19 +253,19 @@ class DiffusionTtsFlat(nn.Module): code_emb) # Everything after this comment is timestep dependent. - x = self.inp_block(x) - pos_emb = self.position_embed(torch.arange(0, x.shape[-1], device=x.device)).unsqueeze(0).repeat(x.shape[0],1,1).permute(0,2,1) - x = x + pos_emb time_emb = self.time_embed(timesteps) code_emb = self.conditioning_timestep_integrator(code_emb, time_emb) + pos_emb = self.position_embed(torch.arange(0, x.shape[-1], device=x.device)).unsqueeze(0).repeat(x.shape[0],1,1).permute(0,2,1) + x = self.inp_block(x) + pos_emb + x = torch.cat([x, F.interpolate(code_emb, size=x.shape[-1], mode='nearest')], dim=1) for i, lyr in enumerate(self.layers): # Do layer drop where applicable. Do not drop first and last layers. - if self.layer_drop > 0 and i != 0 and i != (len(self.layers)-1) and random.random() < self.layer_drop: + if self.training and self.layer_drop > 0 and i != 0 and i != (len(self.layers)-1) and random.random() < self.layer_drop: unused_params.extend(list(lyr.parameters())) else: # First and last blocks will have autocast disabled for improved precision. with autocast(x.device.type, enabled=self.enable_fp16 and i != 0): - x = lyr(x, code_emb, cond_emb, time_emb) + x = lyr(x, time_emb) x = x.float() out = self.out(x) diff --git a/codes/trainer/eval/audio_diffusion_fid.py b/codes/trainer/eval/audio_diffusion_fid.py index d10ce903..3f27e2f5 100644 --- a/codes/trainer/eval/audio_diffusion_fid.py +++ b/codes/trainer/eval/audio_diffusion_fid.py @@ -129,12 +129,15 @@ class AudioDiffusionFid(evaluator.Evaluator): output_size = univnet_mel.shape[-1] aligned_codes_compression_factor = output_size // mel_codes.shape[-1] - padded_size = ceil_multiple(output_size, self.model.alignment_size) - padding_added = padded_size - output_size - padding_needed_for_codes = padding_added // aligned_codes_compression_factor - if padding_needed_for_codes > 0: - mel_codes = F.pad(mel_codes, (0, padding_needed_for_codes)) - output_shape = (1, 100, padded_size) + if hasattr(self.model, 'alignment_size'): + padded_size = ceil_multiple(output_size, self.model.alignment_size) + padding_added = padded_size - output_size + padding_needed_for_codes = padding_added // aligned_codes_compression_factor + if padding_needed_for_codes > 0: + mel_codes = F.pad(mel_codes, (0, padding_needed_for_codes)) + output_shape = (1, 100, padded_size) + else: + output_shape = univnet_mel.shape gen_mel = self.diffuser.p_sample_loop(self.model, output_shape, model_kwargs={'aligned_conditioning': mel_codes, 'conditioning_input': univnet_mel})