train-transducer.yaml

name: "FastConformer-Transducer-BPE"

model:
  sample_rate: 16000
  compute_eval_loss: false
  log_prediction: true 
  rnnt_reduction: 'mean_volume'
  skip_nan_grad: false

  model_defaults:
    enc_hidden: ${model.encoder.d_model}
    pred_hidden: 640
    joint_hidden: 640

  train_ds:
    manifest_filepath: manifest_files/train_manifest.json
    sample_rate: ${model.sample_rate}
    batch_size: 32
    shuffle: true
    num_workers: 4
    pin_memory: true
    max_duration: 16.7
    min_duration: 0.1
    is_tarred: false
    tarred_audio_filepaths: null
    shuffle_n: 2048
    bucketing_strategy: "fully_randomized"
    bucketing_batch_size: null

  validation_ds:
    manifest_filepath: manifest_files/adapt_manifest.json
    sample_rate: ${model.sample_rate}
    batch_size: 32
    shuffle: false
    use_start_end_token: false
    num_workers: 4
    pin_memory: true

  test_ds:
    manifest_filepath: null
    sample_rate: ${model.sample_rate}
    batch_size: 16
    shuffle: false
    use_start_end_token: false
    num_workers: 8
    pin_memory: true

  tokenizer:
    dir: tokenizer
    type: bpe  

  preprocessor:
    _target_: nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor
    sample_rate: ${model.sample_rate}
    normalize: "per_feature"
    window_size: 0.025
    window_stride: 0.01
    window: "hann"
    features: 80
    n_fft: 512
    frame_splicing: 1
    dither: 0.00001
    pad_to: 0

  spec_augment:
    _target_: nemo.collections.asr.modules.SpectrogramAugmentation
    freq_masks: 2
    time_masks: 3
    freq_width: 10
    time_width: 0.05

  encoder:
    _target_: nemo.collections.asr.modules.ConformerEncoder
    feat_in: ${model.preprocessor.features}
    feat_out: -1 
    n_layers: 16
    d_model: 256

    # Sub-sampling parameters
    subsampling: dw_striding 
    subsampling_factor: 8 
    subsampling_conv_channels: 256 
    causal_downsampling: false

    reduction: null 
    reduction_position: null 
    reduction_factor: 1

    # Feed forward module's params
    ff_expansion_factor: 4

    # Multi-headed Attention Module's params
    self_attention_model: rel_pos 
    n_heads: 4 
    att_context_size: [-1, -1] 
    att_context_style: regular 
    xscaling: true 
    untie_biases: true 
    pos_emb_max_len: 5000

    # Convolution module's params
    conv_kernel_size: 9
    conv_norm_type: 'batch_norm' 
    conv_context_size: null

    ### regularization
    dropout: 0.1 
    dropout_pre_encoder: 0.1 
    dropout_emb: 0.0
    dropout_att: 0.1 

    # set to non-zero to enable stochastic depth
    stochastic_depth_drop_prob: 0.0
    stochastic_depth_mode: linear 
    stochastic_depth_start_layer: 1

  decoder:
    _target_: nemo.collections.asr.modules.RNNTDecoder
    normalization_mode: null 
    random_state_sampling: false 
    blank_as_pad: true

    prednet:
      pred_hidden: ${model.model_defaults.pred_hidden}
      pred_rnn_layers: 1
      t_max: null
      dropout: 0.2

  joint:
    _target_: nemo.collections.asr.modules.RNNTJoint
    log_softmax: null
    preserve_memory: false 
    fuse_loss_wer: true
    fused_batch_size: 4

    jointnet:
      joint_hidden: ${model.model_defaults.joint_hidden}
      activation: "relu"
      dropout: 0.2

  decoding:
    strategy: "greedy_batch"

    # greedy strategy config
    greedy:
      max_symbols: 10

    # beam strategy config
    beam:
      beam_size: 2
      return_best_hypothesis: False
      score_norm: true
      tsd_max_sym_exp: 50
      alsd_max_target_len: 2.0

  loss:
    loss_name: "default"

    warprnnt_numba_kwargs:
      fastemit_lambda: 0.0
      clamp: -1.0

  optim:
    name: adamw
    lr: 1e-3
    betas: [0.9, 0.98]
    weight_decay: 1e-3

    sched:
      name: CosineAnnealing
      warmup_steps: 0
      warmup_ratio: null
      min_lr: 5e-4

trainer:
  devices: -1
  num_nodes: 1
  max_epochs: 500
  max_steps: -1
  val_check_interval: 1.0
  accelerator: auto
  strategy: ddp
  accumulate_grad_batches: 1
  gradient_clip_val: 0.0
  precision: 32
  log_every_n_steps: 10
  enable_progress_bar: True
  num_sanity_val_steps: 0
  check_val_every_n_epoch: 1
  sync_batchnorm: true
  enable_checkpointing: False
  logger: false
  benchmark: false


exp_manager:
  exp_dir: null
  name: ${name}
  create_tensorboard_logger: true
  create_checkpoint_callback: true
  checkpoint_callback_params:
    monitor: "val_wer"
    mode: "min"
    save_top_k: 5
    always_save_nemo: True 
  resume_from_checkpoint: null
  resume_if_exists: false
  resume_ignore_no_checkpoint: false

  create_wandb_logger: false
  wandb_logger_kwargs:
    name: null
    project: null