File size: 1,252 Bytes
53c24e9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
paths:
  root_dir: .
  data_dir: ${paths.root_dir}/data/
  log_dir: ${paths.root_dir}/logs/
  output_dir: ${hydra:runtime.output_dir}
  work_dir: ${hydra:runtime.cwd}
data_config:
  name: mitermix/audiosnippets
  streaming: true
  audio_key: mp3
  caption_key: caption
model_config:
  tokenizer_name: openai/whisper-tiny
  model_name: openai/whisper-base
  attn_implementation: flash_attention_2
  name: openai/whisper-tiny
trainer_config:
  tf32: true
  bf16: true
  fp16: false
  dataloader_num_workers: 16
  eval_strategy: 'no'
  eval_steps: null
  save_strategy: steps
  save_steps: 5000
  save_total_limit: 5
  gradient_checkpointing: false
  gradient_checkpointing_kwargs:
    use_reentrant: false
  ddp_find_unused_parameters: false
  logging_steps: 50
  seed: 998244353
  optim: adamw_torch
  adam_beta1: 0.9
  adam_beta2: 0.98
  max_grad_norm: 1.0
  per_device_train_batch_size: 128
  output_dir: ${paths.output_dir}
  report_to: tensorboard
  logging_dir: ${trainer_config.output_dir}/tb
  lr_scheduler_type: cosine
  learning_rate: 0.0001
  warmup_steps: 5000
  max_steps: 50000
  label_smoothing_factor: 0.0
  push_to_hub: true
  hub_model_id: seastar105/whisper-base-emo-speech-caption
  dataloader_prefetch_factor: 4
task_name: tiny-test