You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
PaddleSpeech/examples/mustc/st1/conf/transformer_ru.yaml

90 lines
2.5 KiB

3 years ago
# https://yaml.org/type/float.html
###########################################
# Data #
###########################################
train_manifest: data/manifest.ru.train
dev_manifest: data/manifest.ru.dev
test_manifest: data/manifest.ru.test
###########################################
# Dataloader #
###########################################
vocab_filepath: data/lang_1spm/train_sp.en-ru.ru_bpe8000_units_tc.txt
unit_type: 'spm'
spm_model_prefix: data/lang_1spm/train_sp.en-ru.ru_bpe8000_tc
mean_std_filepath: ""
# preprocess_config: conf/augmentation.json
batch_size: 20
feat_dim: 83
stride_ms: 10.0
window_ms: 25.0
sortagrad: 0 # Feed samples from shortest to longest ; -1: enabled for all epochs, 0: disabled, other: enabled for 'other' epochs
maxlen_in: 512 # if input length > maxlen-in, batchsize is automatically reduced
maxlen_out: 150 # if output length > maxlen-out, batchsize is automatically reduced
minibatches: 0 # for debug
batch_count: auto
batch_bins: 0
batch_frames_in: 0
batch_frames_out: 0
batch_frames_inout: 0
preprocess_config:
num_workers: 0
subsampling_factor: 1
num_encs: 1
############################################
# Network Architecture #
############################################
cmvn_file: None
cmvn_file_type: "json"
# encoder related
encoder: transformer
encoder_conf:
output_size: 256 # dimension of attention
attention_heads: 4
linear_units: 2048 # the number of units of position-wise feed forward
num_blocks: 12 # the number of encoder blocks
dropout_rate: 0.1
positional_dropout_rate: 0.1
attention_dropout_rate: 0.0
input_layer: conv2d # encoder input type, you can chose conv2d, conv2d6 and conv2d8
normalize_before: true
# decoder related
decoder: transformer
decoder_conf:
attention_heads: 4
linear_units: 2048
num_blocks: 6
dropout_rate: 0.1
positional_dropout_rate: 0.1
self_attention_dropout_rate: 0.0
src_attention_dropout_rate: 0.0
# hybrid CTC/attention
model_conf:
asr_weight: 0.0
ctc_weight: 0.0
lsm_weight: 0.1 # label smoothing option
length_normalized_loss: false
###########################################
# Training #
###########################################
n_epoch: 40
accum_grad: 2
global_grad_clip: 5.0
optim: adam
optim_conf:
lr: 2.5
weight_decay: 0.
scheduler: noam
scheduler_conf:
warmup_steps: 25000
lr_decay: 1.0
log_interval: 50
checkpoint:
kbest_n: 50
latest_n: 5