Commit ad46064b by xuchen

Modify the egs scripts of mustc

parent c4f60362
train-subset: train_asr
valid-subset: dev_asr
max-epoch: 50
max-update: 100000
num-workers: 8
patience: 10
no-progress-bar: True
log-interval: 100
seed: 1
report-accuracy: True
#load-params:
#load-pretrained-encoder-from:
arch: s2t_transformer_s
share-decoder-input-output-embed: True
optimizer: adam
clip-norm: 10.0
lr-scheduler: inverse_sqrt
warmup-init-lr: 1e-7
warmup-updates: 10000
lr: 1e-3
#adam_betas: (0.9,0.98)
criterion: label_smoothed_cross_entropy
label_smoothing: 0.1
conv-kernel-sizes: 5,5
conv-channels: 1024
dropout: 0.1
activation-fn: relu
encoder-embed-dim: 256
encoder-ffn-embed-dim: 2048
encoder-layers: 12
decoder-layers: 6
encoder-attention-heads: 4
#decoder-embed-dim: 256
#decoder-ffn-embed-dim: 2048
#decoder-attention-heads: 4
#attention-dropout: 0.1
#activation-dropout: 0.1
......@@ -29,7 +29,8 @@ pwd_dir=$PWD
# dataset
src_lang=en
lang=${src_lang}
tgt_lang=de
lang=${src_lang}-${tgt_lang}
dataset=mustc
task=speech_to_text
......@@ -124,7 +125,7 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
cmd="python3 -u ${root_dir}/fairseq_cli/train.py
$data_dir/$lang
--config-yaml ${data_config}
--config-yaml ${data_config}
--train-config ${train_config}
--task speech_to_text
--max-tokens ${max_tokens}
......
Markdown 格式
0%
您添加了 0 到此讨论。请谨慎行事。
请先完成此评论的编辑!
注册 或者 后发表评论