forked from jmoso13/jukebox-diffusion
-
Notifications
You must be signed in to change notification settings - Fork 0
/
jbdiff-sample-v1.yaml
80 lines (80 loc) · 3.52 KB
/
jbdiff-sample-v1.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
sampling:
diffusion:
dd:
num_steps: 150
init_strength: 0.28
ckpt_loc: "./epoch=2125-step=218000.ckpt"
xfade_samples: 1536
xfade_style: "constant-power"
0:
num_steps: 20
embedding_strength: 1.3
init_strength: 0.5
ckpt_loc: "./epoch=543-step=705000.ckpt"
1:
num_steps: 70
embedding_strength: 2.0
init_strength: 0.67
ckpt_loc: "./epoch=1404-step=455000.ckpt"
2:
num_steps: 250
embedding_strength: 4.0
ckpt_loc: "./epoch=4938-step=400000_vqvae_add.ckpt"
model:
vqvae:
batch_size: 32
aug_shift: True
base_tokens: 768
context_mult: 2
diffusion:
2: #Level 2 is bottom level of vqvae
net_t: audio_diffusion_pytorch.UNetV0
in_channels: 64
channels: [64, 128, 128, 256, 256, 512, 512, 1024, 1024]
factors: [1, 2, 1, 2, 1, 2, 1, 2, 1]
items: [1, 3, 3, 3, 3, 3, 3, 3, 1]
attentions: [0, 0, 0, 0, 1, 0, 1, 0, 0]
cross_attentions: [0, 1, 0, 1, 0, 1, 0, 1, 1]
attention_heads: 8
attention_features: 64
diffusion_t: audio_diffusion_pytorch.VDiffusion
sampler_t: audio_diffusion_pytorch.VSampler
embedding_features: 64
use_embedding_cfg: True
# embedding_max_length: context_mult*base_tokens
resnet_dilation_factor: 3
resnet_dropout_rate: 0.05
1: #Level 1 is middle level of vqvae
net_t: audio_diffusion_pytorch.UNetV0
in_channels: 64
channels: [64, 128, 128, 256, 256, 512, 512, 1024, 1024]
factors: [1, 2, 1, 2, 1, 2, 1, 2, 1]
items: [1, 3, 3, 3, 3, 3, 3, 3, 1]
attentions: [0, 0, 0, 0, 1, 0, 1, 0, 0]
cross_attentions: [0, 1, 0, 1, 0, 1, 0, 1, 1]
attention_heads: 8
attention_features: 64
diffusion_t: audio_diffusion_pytorch.VDiffusion
sampler_t: audio_diffusion_pytorch.VSampler
embedding_features: 64
use_embedding_cfg: True
# embedding_max_length: context_mult*base_tokens
resnet_dilation_factor: 3
resnet_dropout_rate: 0.00
0: #Level 0 is top level of vqvae
net_t: audio_diffusion_pytorch.UNetV0
in_channels: 64
channels: [64, 128, 128, 256, 256, 512, 512, 1024, 1024]
factors: [1, 2, 1, 2, 1, 2, 1, 2, 1]
items: [1, 3, 3, 3, 3, 3, 3, 3, 1]
attentions: [0, 0, 0, 0, 1, 0, 1, 0, 0]
cross_attentions: [0, 1, 0, 1, 0, 1, 0, 1, 1]
attention_heads: 8
attention_features: 64
diffusion_t: audio_diffusion_pytorch.VDiffusion
sampler_t: audio_diffusion_pytorch.VSampler
embedding_features: 64
use_embedding_cfg: True
# embedding_max_length: context_mult*base_tokens
resnet_dilation_factor: 3
resnet_dropout_rate: 0.00