NikkeDoy/SISU
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge new-arch, because it has proven to give the best results #1

Merged
NikkeDoy merged 14 commits from new-arch into main 2025-04-30 23:47:41 +03:00
Conversation 0 Commits 14 Files Changed 8 +9 -10
3 changed files with 9 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit fb7b624c87 - Show all commits

9
discriminator.py
View File

@ -13,7 +13,7 @@ def discriminator_block(in_channels, out_channels, kernel_size=3, stride=1, dila
class SISUDiscriminator(nn.Module):
def __init__(self):
super(SISUDiscriminator, self).__init__()
layers = 32 # Increased base layer count
layers = 4 # Increased base layer count
self.model = nn.Sequential(
# Initial Convolution
discriminator_block(1, layers, kernel_size=7, stride=2, dilation=1), # Downsample
@ -21,10 +21,9 @@ class SISUDiscriminator(nn.Module):
# Core Discriminator Blocks with varied kernels and dilations
discriminator_block(layers, layers * 2, kernel_size=5, stride=2, dilation=1), # Downsample
discriminator_block(layers * 2, layers * 4, kernel_size=5, dilation=4),
discriminator_block(layers * 4, layers * 8, kernel_size=5, dilation=16),
discriminator_block(layers * 8, layers * 4, kernel_size=3, dilation=8),
discriminator_block(layers * 4, layers * 2, kernel_size=3, dilation=2),
discriminator_block(layers * 2, layers, kernel_size=5, dilation=1),
discriminator_block(layers * 4, layers * 4, kernel_size=5, dilation=16),
discriminator_block(layers * 4, layers * 2, kernel_size=3, dilation=8),
discriminator_block(layers * 2, layers, kernel_size=3, dilation=1),
# Final Convolution
discriminator_block(layers, 1, kernel_size=3, stride=1),
)

8
generator.py
View File

@ -10,7 +10,7 @@ def conv_block(in_channels, out_channels, kernel_size=3, dilation=1):
class SISUGenerator(nn.Module):
def __init__(self):
super(SISUGenerator, self).__init__()
layer = 32 # Increased base layer count
layer = 4 # Increased base layer count
self.conv1 = nn.Sequential(
nn.Conv1d(1, layer, kernel_size=7, padding=3),
nn.BatchNorm1d(layer),
@ -18,9 +18,9 @@ class SISUGenerator(nn.Module):
)
self.conv_blocks = nn.Sequential(
conv_block(layer, layer, kernel_size=3, dilation=1), # Local details
conv_block(layer, layer*4, kernel_size=5, dilation=2), # Local Context
conv_block(layer*4, layer*4, kernel_size=3, dilation=16), # Longer range dependencies
conv_block(layer*4, layer*2, kernel_size=5, dilation=8), # Wider context
conv_block(layer, layer*2, kernel_size=5, dilation=2), # Local Context
conv_block(layer*2, layer*2, kernel_size=3, dilation=16), # Longer range dependencies
conv_block(layer*2, layer*2, kernel_size=5, dilation=8), # Wider context
conv_block(layer*2, layer, kernel_size=5, dilation=2), # Local Context
conv_block(layer, layer, kernel_size=3, dilation=1), # Local details
)

2
training.py
View File

@ -85,7 +85,7 @@ dataset = AudioDataset(dataset_dir)
# ========= SINGLE =========
train_data_loader = DataLoader(dataset, batch_size=1, shuffle=True)
train_data_loader = DataLoader(dataset, batch_size=16, shuffle=True)
# Initialize models and move them to device
generator = SISUGenerator()