From 219c2482ab755fbd4e49dfbdee91bf1a8a4ec49a Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期五, 19 五月 2023 11:33:27 +0800
Subject: [PATCH] websocket 2pass bugfix
---
egs/aishell/data2vec_transformer_finetune/run.sh | 31 +++++++++++++++++++------------
1 files changed, 19 insertions(+), 12 deletions(-)
diff --git a/egs/aishell/data2vec_transformer_finetune/run.sh b/egs/aishell/data2vec_transformer_finetune/run.sh
index e040290..af0b8c1 100755
--- a/egs/aishell/data2vec_transformer_finetune/run.sh
+++ b/egs/aishell/data2vec_transformer_finetune/run.sh
@@ -8,7 +8,7 @@
count=1
gpu_inference=true # Whether to perform gpu decoding, set false for cpu decoding
# for gpu decoding, inference_nj=ngpu*njob; for cpu decoding, inference_nj=njob
-njob=1
+njob=5
train_cmd=utils/run.pl
infer_cmd=utils/run.pl
@@ -20,15 +20,15 @@
type=sound
scp=wav.scp
speed_perturb="0.9 1.0 1.1"
-stage=3
-stop_stage=4
+stage=0
+stop_stage=5
# feature configuration
feats_dim=80
nj=64
# data
-raw_data=
+raw_data=../raw_data
data_url=www.openslr.org/resources/33
# exp tag
@@ -52,8 +52,8 @@
asr_config=conf/train_asr_transformer_12e_6d_3072_768.yaml
model_dir="baseline_$(basename "${asr_config}" .yaml)_${lang}_${token_type}_${tag}"
-inference_config=conf/decode_asr_transformer_noctc_1best.yaml
-inference_asr_model=valid.acc.ave_10best.pb
+inference_config=conf/decode_asr_transformer.yaml
+inference_asr_model=valid.cer_ctc.ave_10best.pb
# you can set gpu num for decoding here
gpuid_list=$CUDA_VISIBLE_DEVICES # set gpus for decoding, the same as training stage by default
@@ -106,11 +106,17 @@
echo "<unk>" >> ${token_list}
fi
-# Training Stage
+# LM Training Stage
world_size=$gpu_num # run on one machine
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
- echo "stage 3: Training"
- python utils/download_model.py --model_name ${model_name} # download pretrained model on ModelScope
+ echo "stage 3: LM Training"
+fi
+
+# ASR Training Stage
+world_size=$gpu_num # run on one machine
+if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
+ echo "stage 4: ASR Training"
+ python utils/download_model.py --model_name ${model_name} # download pretrained model on ModelScope
mkdir -p ${exp_dir}/exp/${model_dir}
mkdir -p ${exp_dir}/exp/${model_dir}/log
INIT_FILE=${exp_dir}/exp/${model_dir}/ddp_init
@@ -133,6 +139,7 @@
--data_dir ${feats_dir}/data \
--train_set ${train_set} \
--valid_set ${valid_set} \
+ --data_file_names "wav.scp,text" \
--init_param ${init_param} \
--cmvn_file ${feats_dir}/data/${train_set}/cmvn/cmvn.mvn \
--speed_perturb ${speed_perturb} \
@@ -151,8 +158,8 @@
fi
# Testing Stage
-if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
- echo "stage 4: Inference"
+if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
+ echo "stage 5: Inference"
for dset in ${test_sets}; do
asr_exp=${exp_dir}/exp/${model_dir}
inference_tag="$(basename "${inference_config}" .yaml)"
@@ -189,7 +196,7 @@
--asr_train_config "${asr_exp}"/config.yaml \
--asr_model_file "${asr_exp}"/"${inference_asr_model}" \
--output_dir "${_logdir}"/output.JOB \
- --mode paraformer \
+ --mode asr \
${_opts}
for f in token token_int score text; do
--
Gitblit v1.9.1