File size: 1,211 Bytes
4725118
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# @package __global__

# This is the training loop solver
# for the base MusicGen model (text-to-music)
defaults:
  - musicgen/default
  - /model: lm/musicgen_lm
  - _self_

autocast: true
autocast_dtype: float16

# EnCodec large trained on mono-channel music audio sampled at 32khz
# with a total stride of 640 leading to 50 frames/s.
# rvq.n_q=4, rvq.bins=2048, no quantization dropout
# (transformer_lm card and n_q must be compatible)
compression_model_checkpoint: //pretrained/facebook/encodec_32khz

channels: 1
sample_rate: 32000

deadlock:
  use: true  # deadlock detection

dataset:
  batch_size: 8  # 4 GPUs(3090)
  num_workers: 8
  segment_duration: 30
  sample_on_weight: false  # Uniform sampling all the way
  sample_on_duration: false  # Uniform sampling all the way
  valid:
    num_samples: 4

generate:
  lm:
    use_sampling: true
    top_k: 250
    top_p: 0.0

checkpoint:
  save_last: true
  save_every: 25
  keep_every_states: null

optim:
  epochs: 100
  optimizer: dadam
  lr: 1.0
  max_norm: 1.0
  ema:
    use: false
    updates: 10
    device: cuda

logging:
  log_tensorboard: true

schedule:
  lr_scheduler: cosine
  cosine:
    warmup: 5
    lr_min_ratio: 0.0
    cycle_length: 1.0