-
Notifications
You must be signed in to change notification settings - Fork 22
/
Copy pathbase_youtube_6.yaml
169 lines (158 loc) · 6.78 KB
/
base_youtube_6.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
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
# @package _global_
# 106M params
sampling_rate: 48000
length: 524288
channels: 2
log_every_n_steps: 1000
model:
_target_: main.module_base.Model
learning_rate: 1e-4
beta1: 0.9
beta2: 0.99
ema_beta: 0.9999
ema_power: 0.75
model:
_target_: audio_diffusion_pytorch.AudioDiffusionModel
in_channels: ${channels}
channels: 128
patch_factor: 16
patch_blocks: 1
resnet_groups: 8
kernel_multiplier_downsample: 2
multipliers: [1, 2, 4, 4, 4, 4, 4]
factors: [4, 4, 4, 2, 2, 2]
num_blocks: [2, 2, 2, 2, 2, 2]
attentions: [0, 0, 0, 1, 1, 1, 1]
attention_heads: 8
attention_features: 64
attention_multiplier: 2
use_nearest_upsample: False
use_skip_scale: True
use_magnitude_channels: True
diffusion_sigma_distribution:
_target_: audio_diffusion_pytorch.LogNormalDistribution
mean: -3.0
std: 1.0
diffusion_sigma_data: 0.2
diffusion_dynamic_threshold: 0.0
datamodule:
_target_: main.module_base.Datamodule
dataset:
_target_: audio_data_pytorch.YoutubeDataset
urls:
- https://www.youtube.com/watch?v=F1gIsoIVQG8 # 2h Kiasmos
- https://www.youtube.com/watch?v=KjgluLOMa0k # 1h Techno Mix
- https://www.youtube.com/watch?v=c_iRx2Un07k # 1h Kygo Piano
- https://www.youtube.com/watch?v=fWRISvgAygU # 2h Chillstep
- https://www.youtube.com/watch?v=aJoo79OwZEI # 1h Drum and Bass
- https://www.youtube.com/watch?v=vnKKNZLVh2Q # 1h French 79
- https://www.youtube.com/watch?v=5gcyzjCLH90 # 1h Andrew Langdon
- https://www.youtube.com/watch?v=FFfdyV8gnWk # 1h City of Gamers
- https://www.youtube.com/watch?v=eqhQwgO03OE # 1h Lofi beats
- https://www.youtube.com/watch?v=_dEw2zQ7xcE # 1h GHØSTS
- https://www.youtube.com/watch?v=W_cu3mS5UU4 # 1h Cyberpunk 2077
- https://www.youtube.com/watch?v=0mAzlQwIPQI # 2h Dark Techno
- https://www.youtube.com/watch?v=k3WkJq478To # 2.5h Synthwave Mix
- https://www.youtube.com/watch?v=Et0bFcvSImY # 1.5h Worakls, N'to , Joachim Pastor Mix
- https://www.youtube.com/watch?v=k3WkJq478To # 2.5h Synthwave Mix
- https://www.youtube.com/watch?v=yGsGdhe0pj8 # Piano Solo, 2h
- https://www.youtube.com/watch?v=PJL_mVgT0Ao # Piano Solo 6h
- https://www.youtube.com/watch?v=GuCqvjcZTKM # Einaudi, 2h
- https://www.youtube.com/watch?v=FccYUW91an8 # Mendelssohn, 4h
- https://www.youtube.com/watch?v=CuU9q2VKOyc # Baroque, 4h
- https://www.youtube.com/watch?v=7JynrZsIrgo # Chopin, 3h
- https://www.youtube.com/watch?v=_ioc6sdgugo # Bach, 1.2h
- https://www.youtube.com/watch?v=EhO_MrRfftU # 4h Mix piano
- https://www.youtube.com/watch?v=c_iRx2Un07k # 1h Kygo
- https://www.youtube.com/watch?v=cGYyOY4XaFs # 2.5h Mix
- https://www.youtube.com/watch?v=GDnU6nbL1uQ # 1.5h Avicii
- https://www.youtube.com/watch?v=_dmOgDlWAkU # 1h Einaudi Piano Only
- https://www.youtube.com/watch?v=E7EOjkGVmyo # 1h Jacobs
- https://www.youtube.com/watch?v=5yJOf1JJ5b8 # 1h Relax
- https://www.youtube.com/watch?v=4fezP875xOQ # 4h Chopin
- https://www.youtube.com/watch?v=5-Y35hjZ06E # 4h Schubert
- https://www.youtube.com/watch?v=cMzyLBuFm4A # 2h Yiurma
- https://www.youtube.com/watch?v=7UtIg4dOe84 # 3h, Hans Zimmer
- https://www.youtube.com/watch?v=DWDZSqBHW5g # 2h, Ludovico Einaudi
- https://www.youtube.com/watch?v=aVMkvCTT_yg # 1.5h, Ludwig Göransson
- https://www.youtube.com/watch?v=Et0bFcvSImY # 1.5h, Worakls, N'to , Joachim Pastor Mix
- https://www.youtube.com/watch?v=gThS-KfIxOs # 2h, Beatles
- https://www.youtube.com/watch?v=3jgH5weXYwA # 3h, 2022 Mix
- https://www.youtube.com/watch?v=6JQm5aSjX6g # 2h, Bach Mix
- https://www.youtube.com/watch?v=k3WkJq478To # 2.5h Synthwave Mix
- https://www.youtube.com/watch?v=_3A4JW9RQWM # 1.5h Rock Mix
- https://www.youtube.com/watch?v=jgpJVI3tDbY # 3.5h Classical Mix
- https://www.youtube.com/watch?v=KjgluLOMa0k # 1h Techno Mix
- https://www.youtube.com/watch?v=zK3uaiVckEk # 1h Trap Mix
- https://www.youtube.com/watch?v=JQ1txLdu6qg # 1h Dubstep Mix
- https://www.youtube.com/watch?v=m8F8v2t5VcQ # 1.5h EDM
- https://www.youtube.com/watch?v=fSH9sWGzARA # 4h Techno Mix
- https://www.youtube.com/watch?v=VTVC4weBFUA # 10h Jazz
- https://www.youtube.com/watch?v=7sTlQiBtB1A # 11h Metal
- https://www.youtube.com/watch?v=26nsBfLXwSQ # 2h Rock
- https://www.youtube.com/watch?v=EbHxWU52ZH0 # 2.5h 90s
- https://www.youtube.com/watch?v=1L2RgI2G0qY # 2.4h 2000s
- https://www.youtube.com/watch?v=1L2RgI2G0qY # 2h 80s
root: ${data_dir}
crop_length: 60 # seconds crops
transforms:
_target_: audio_data_pytorch.AllTransform
source_rate: 48000
target_rate: ${sampling_rate}
random_crop_size: ${length}
#loudness: -20
val_split: 0.0005
batch_size: 32
num_workers: 8
pin_memory: True
callbacks:
rich_progress_bar:
_target_: pytorch_lightning.callbacks.RichProgressBar
model_checkpoint:
_target_: pytorch_lightning.callbacks.ModelCheckpoint
monitor: "valid_loss" # name of the logged metric which determines when model is improving
save_top_k: 1 # save k best models (determined by above metric)
save_last: True # additionaly always save model from last epoch
mode: "min" # can be "max" or "min"
verbose: False
dirpath: ${logs_dir}/ckpts/${now:%Y-%m-%d-%H-%M-%S}
filename: '{epoch:02d}-{valid_loss:.3f}'
model_summary:
_target_: pytorch_lightning.callbacks.RichModelSummary
max_depth: 2
audio_samples_logger:
_target_: main.module_base.SampleLogger
num_items: 3
channels: ${channels}
sampling_rate: ${sampling_rate}
length: ${length}
sampling_steps: [3,5,10,25,50,100]
diffusion_sampler:
_target_: audio_diffusion_pytorch.ADPM2Sampler
rho: 1.0
diffusion_schedule:
_target_: audio_diffusion_pytorch.KarrasSchedule
sigma_min: 0.0001
sigma_max: 3.0
rho: 9.0
loggers:
wandb:
_target_: pytorch_lightning.loggers.wandb.WandbLogger
project: ${oc.env:WANDB_PROJECT}
entity: ${oc.env:WANDB_ENTITY}
# offline: False # set True to store all logs only locally
job_type: "train"
group: ""
save_dir: ${logs_dir}
trainer:
_target_: pytorch_lightning.Trainer
gpus: 0 # Set `1` to train on GPU, `0` to train on CPU only, and `-1` to train on all GPUs, default `0`
precision: 32 # Precision used for tensors, default `32`
accelerator: null # `ddp` GPUs train individually and sync gradients, default `None`
min_epochs: 0
max_epochs: -1
enable_model_summary: False
log_every_n_steps: 1 # Logs metrics every N batches
check_val_every_n_epoch: null
val_check_interval: ${log_every_n_steps}
accumulate_grad_batches: 1