Hugging Face's logo

Spaces:
H-Liu1997
/
EMAGE
Running on T4

App Files Community
2
EMAGE
File size: 1,400 Bytes 
b03a8f2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
wandb_project: 'EMAGE++'
exp_name: 'debug'
debug: False
wandb_entity: ''
wandb_key: ""
# wandb_log_dir: '/content/outputs/wandb'
output_dir: "../outputs/"
log_period: 1
seed: 42
resume_from_checkpoint: null
test: False


data:
  name_pyfile: "datasets.beat2"
  class_name: "BEAT2Dataset"
  train_bs: 64
  meta_paths:
    - "./datasets/data_json/beat2_s20_l128_speaker2.json"
  test_meta_paths: 
    - "./datasets/data_json/beat2_s20_l128_speaker2.json"
  pose_norm: False
  pose_length: 128
  stride: 20
  test_length: 128

model:
  name_pyfile: "models.camn_audio.modeling_camn_audio"
  class_name: "CamnAudioModel"
  pose_fps: 15
  motion_f: 256
  pose_dims: 258
  pose_rep: "smplx"
  body_dims: 78
  hands_dims: 180
  audio_rep: wave16k
  audio_sr: 16000
  audio_fps: 16000
  audio_norm: False
  audio_f: 128
  speaker_f: 16
  speaker_dims: 1
  hidden_size: 512
  n_layer: 4
  dropout_prob: 0.1
  seed_frames: 4 
  joint_mask: "local_upper"
 
validation:
  validation_steps: 500
  test_steps: 500
  visualization: False
  evaluation: False
  wandb: False

solver:
  gradient_accumulation_steps: 1
  gradient_checkpointing: False 
  max_train_steps: 100000
  max_grad_norm: 0.0
  # lr
  learning_rate: 3e-4
  scale_lr: False 
  lr_warmup_steps: 0
  lr_scheduler: 'constant'
  # optimizer
  use_8bit_adam: False 
  adam_beta1: 0.9
  adam_beta2: 0.999
  adam_weight_decay:  0.0
  adam_epsilon: 1.0e-8