runner:
  total_steps: 200000
  gradient_clipping: 1
  gradient_accumulate_steps: 1

  log_step: 100
  eval_step: 2000
  save_step: 500
  max_keep: 1
  eval_dataloaders:
    - dev

optimizer:
  name: TorchOptim
  torch_optim_name: Adam
  lr: 1.0e-4

# comment the whole scheduler config block
# to disable learning rate scheduling
# scheduler:
#   name: linear_schedule_with_warmup
#   num_warmup_steps: 1400

specaug:
  adaptive: false
  adaptive_number_ratio: 0.04
  adaptive_size_ratio: 0.04
  max_n_time_masks: 20
  apply_time_warp: true
  apply_time_mask: true
  apply_freq_mask: true
  time_warp_window: 5
  time_mask_width_range: [0, 40]
  freq_mask_width_range: [0, 50]
  num_freq_mask: 4
  num_time_mask: 2

downstream_expert:
  corpus:                                 
    name: 'snips'                   # Specify corpus
    path: '/path/to/SNIPS'          # Path to raw LibriSpeech dataset

    train: ['train']                # Name of data splits to be used as training set
    dev: ['valid']                    # Name of data splits to be used as validation set
    test: ['test']

    bucketing: True                       # Enable/Disable bucketing 
    batch_size: 32
    train_speakers: 
    # - Aditi
    # - Amy
    # - Brian
    # - Emma
    # - Geraint
    - Ivy
    - Joanna
    - Joey
    - Justin
    - Kendra
    - Kimberly
    - Matthew
    # - Nicole
    # - Raveena
    # - Russell
    - Salli
    valid_speakers:
    - Aditi
    - Amy
    # - Brian
    # - Emma
    - Geraint
    # - Ivy
    # - Joanna
    # - Joey
    # - Justin
    # - Kendra
    # - Kimberly
    # - Matthew
    - Nicole
    # - Raveena
    # - Russell
    # - Salli
    test_speakers:
    # - Aditi
    # - Amy
    - Brian
    - Emma
    # - Geraint
    # - Ivy
    # - Joanna
    # - Joey
    # - Justin
    # - Kendra
    # - Kimberly
    # - Matthew
    # - Nicole
    - Raveena
    - Russell
    # - Salli
    num_workers: 12
  
  text:
    mode: 'character-slot'                       # 'character'/'word'/'subword'
    vocab_file: 'downstream/ctc/vocab/character.txt'
    slots_file: '/path/to/SNIPS/slots.txt'
  
  model:
    project_dim: 1024
    zero_infinity: True

    select: RNNs
    Wav2Letter:
      total_rate: 320
    RNNs:
      total_rate: -1
      module: 'LSTM'                        # 'LSTM'/'GRU'
      bidirection: True
      dim: [1024, 1024]
      dropout: [0.2, 0.2]
      layer_norm: [False, False]
      proj: [False, False]              # Linear projection + Tanh after each rnn layer
      sample_rate: [1, 1]
      sample_style: 'concat'                  # 'drop'/'concat'

  save_best_on:
    - dev

  metric_higher_better: True
  metric:  # The first metric will be used to save checkpoint
    - slot_type_f1
    - slot_value_cer
    - slot_value_wer
    - slot_edit_f1_full
    - slot_edit_f1_part
    - wer
    - cer