FunASR/examples/industrial_data_pretraining/paraformer-large/conf/finetune.yaml
2023-12-11 13:42:40 +08:00

120 lines
2.2 KiB
YAML

# network architecture
model: funasr.cli.models.paraformer:Paraformer
model_conf:
ctc_weight: 0.0
lsm_weight: 0.1
length_normalized_loss: true
predictor_weight: 1.0
predictor_bias: 1
sampling_ratio: 0.75
# encoder
encoder: sanm
encoder_conf:
output_size: 512
attention_heads: 4
linear_units: 2048
num_blocks: 50
dropout_rate: 0.1
positional_dropout_rate: 0.1
attention_dropout_rate: 0.1
input_layer: pe
pos_enc_class: SinusoidalPositionEncoder
normalize_before: true
kernel_size: 11
sanm_shfit: 0
selfattention_layer_type: sanm
# decoder
decoder: paraformer_decoder_sanm
decoder_conf:
attention_heads: 4
linear_units: 2048
num_blocks: 16
dropout_rate: 0.1
positional_dropout_rate: 0.1
self_attention_dropout_rate: 0.1
src_attention_dropout_rate: 0.1
att_layer_num: 16
kernel_size: 11
sanm_shfit: 0
predictor: cif_predictor_v2
predictor_conf:
idim: 512
threshold: 1.0
l_order: 1
r_order: 1
tail_threshold: 0.45
# frontend related
frontend: wav_frontend
frontend_conf:
fs: 16000
window: hamming
n_mels: 80
frame_length: 25
frame_shift: 10
lfr_m: 7
lfr_n: 6
specaug: specaug_lfr
specaug_conf:
apply_time_warp: false
time_warp_window: 5
time_warp_mode: bicubic
apply_freq_mask: true
freq_mask_width_range:
- 0
- 30
lfr_rate: 6
num_freq_mask: 1
apply_time_mask: true
time_mask_width_range:
- 0
- 12
num_time_mask: 1
train_conf:
accum_grad: 1
grad_clip: 5
max_epoch: 150
val_scheduler_criterion:
- valid
- acc
best_model_criterion:
- - valid
- acc
- max
keep_nbest_models: 10
log_interval: 50
optim: adam
optim_conf:
lr: 0.0005
scheduler: warmuplr
scheduler_conf:
warmup_steps: 30000
dataset_conf:
data_names: speech,text
data_types: sound,text
shuffle: True
shuffle_conf:
shuffle_size: 2048
sort_size: 500
batch_conf:
batch_type: example
batch_size: 2
num_workers: 8
split_with_space: true
input_size: 560
ctc_conf:
dropout_rate: 0.0
ctc_type: builtin
reduce: true
ignore_nan_grad: true
normalize: null