train.yaml

accum_grad: 1
cmvn_file: exp/wenet_efficient_conformer_aishell_v2/global_cmvn
dataset_conf:
  batch_conf:
    batch_size: 16
    batch_type: static
  fbank_conf:
    dither: 1.0
    frame_length: 25
    frame_shift: 10
    num_mel_bins: 80
  filter_conf:
    max_length: 40960
    max_output_input_ratio: 0.1
    min_length: 0
    min_output_input_ratio: 0.005
    token_max_length: 200
    token_min_length: 1
  resample_conf:
    resample_rate: 16000
  shuffle: true
  shuffle_conf:
    shuffle_size: 1500
  sort: true
  sort_conf:
    sort_size: 500
  spec_aug: true
  spec_aug_conf:
    max_f: 10
    max_t: 50
    num_f_mask: 2
    num_t_mask: 2
  spec_sub: true
  spec_sub_conf:
    max_t: 30
    num_t_sub: 3
  spec_trim: false
  spec_trim_conf:
    max_t: 50
  speed_perturb: true
decoder: bitransformer
decoder_conf:
  attention_heads: 8
  dropout_rate: 0.1
  linear_units: 2048
  num_blocks: 3
  positional_dropout_rate: 0.1
  r_num_blocks: 3
  self_attention_dropout_rate: 0.1
  src_attention_dropout_rate: 0.1
encoder: efficientConformer
encoder_conf:
  activation_type: swish
  attention_dropout_rate: 0.1
  attention_heads: 8
  causal: false
  cnn_module_kernel: 15
  cnn_module_norm: layer_norm
  dropout_rate: 0.1
  efficient_conf:
    group_layer_idx:
    - 3
    - 7
    group_size: 3
    stride:
    - 2
    - 2
    stride_kernel: false
    stride_layer_idx:
    - 3
    - 7
  input_layer: conv2d2
  linear_units: 2048
  normalize_before: true
  num_blocks: 12
  output_size: 256
  pos_enc_layer_type: rel_pos
  positional_dropout_rate: 0.1
  use_cnn_module: true
  use_dynamic_chunk: true
  use_dynamic_left_chunk: false
grad_clip: 5
input_dim: 80
is_json_cmvn: true
log_interval: 100
max_epoch: 200
model_conf:
  ctc_weight: 0.3
  length_normalized_loss: false
  lsm_weight: 0.1
  reverse_weight: 0.3
optim: adam
optim_conf:
  lr: 0.001
output_dim: 4233
scheduler: warmuplr
scheduler_conf:
  warmup_steps: 25000