mrq/DL-Art-School
1
3
Fork 5
Code Issues Pull Requests Packages Projects Releases Wiki Activity

better downsampler

Browse Source
This commit is contained in:
James Betker 2022-06-16 15:19:08 -06:00
parent 28d95e3141
commit 41abf776d9
1 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
codes/models/audio/music/gpt_music2.py
View File

@ -18,17 +18,17 @@ class UpperEncoder(nn.Module):
super().__init__() super().__init__()
attn = [] attn = []
def edim(m): def edim(m):
dd = max(hidden_dim // m, 128, spec_dim) dd = min(spec_dim + m * 128, hidden_dim)
return ceil_multiple(dd, 8) return ceil_multiple(dd, 8)
self.downsampler = nn.Sequential( self.downsampler = nn.Sequential(
ResBlock(spec_dim, out_channels=edim(6), use_conv=True, dims=1, down=True), ResBlock(spec_dim, out_channels=edim(1), use_conv=True, dims=1, down=True),
ResBlock(edim(6), out_channels=edim(5), use_conv=True, dims=1, down=True), ResBlock(edim(1), out_channels=edim(2), use_conv=True, dims=1, down=True),
ResBlock(edim(5), out_channels=edim(4), use_conv=True, dims=1, down=True), ResBlock(edim(2), out_channels=edim(3), use_conv=True, dims=1, down=True),
ResBlock(edim(4), out_channels=edim(3), use_conv=True, dims=1, down=True), ResBlock(edim(3), out_channels=edim(4), use_conv=True, dims=1, down=True),
ResBlock(edim(3), out_channels=edim(3), use_conv=True, dims=1), ResBlock(edim(4), out_channels=edim(4), use_conv=True, dims=1),
ResBlock(edim(3), out_channels=edim(2), use_conv=True, dims=1, down=True), ResBlock(edim(4), out_channels=edim(5), use_conv=True, dims=1, down=True),
ResBlock(edim(2), out_channels=edim(2), use_conv=True, dims=1), ResBlock(edim(5), out_channels=edim(5), use_conv=True, dims=1),
ResBlock(edim(2), out_channels=hidden_dim, use_conv=True, dims=1, down=True)) ResBlock(edim(5), out_channels=hidden_dim, use_conv=True, dims=1, down=True))
self.encoder = nn.Sequential( self.encoder = nn.Sequential(
AttentionBlock(hidden_dim, 4, do_activation=True), AttentionBlock(hidden_dim, 4, do_activation=True),
ResBlock(hidden_dim, out_channels=hidden_dim, use_conv=True, dims=1), ResBlock(hidden_dim, out_channels=hidden_dim, use_conv=True, dims=1),