-
modelscope版本:1.15.0
python版本:3.8.0
运行CTC的demo程序:
```
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
kws = pipeline(
Tasks.keyword_spotting,
…
-
now DFSMN is use single GPU ,when ali privode muti gpu version
I use BMUF method for muti GPU ,but wer is not better than single GPU.
-
运行命令:local/nnet/run_fsmn_ivector.sh DFSMN_S
之后提示:
cp: cannot stat 'data/train_960_cleaned': No such file or directory
-
My System info:
Ubuntu: 16.04
g++: 4.8.5
with no GPU
Quest:
after typing "local/nnet/run_fsmn_ivector.sh DFSMN_S", it indeed ran for a while, but then "local/nnet/run_fsmn_ivector.sh: line 27: sy…
-
-
您好,在训练过程中有用什么训练策略吗?比如调整音量、速度等?
-
When to extract the fbank feature, the fbank.cfg is not in conf dirs, so how can I get it?
-
在不,你这个代码能用么,utils下面没有tools,也没有提到数据集,这个vacab_file也没提供,能解决下么?
-
modelscope版本:1.15.0
运行CTC的demo程序:
```
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
kws = pipeline(
Tasks.keyword_spotting,
model='damo/spe…
-
**介绍:**
我原先安装的modelscope版本为1.1.1的,但是在网上看到说可以降低一下就又安装了1.10.0版本的,还是报这个信息,我在打包前使用官方的验证内置模型代码是没问题的,使用pyinstaller打包起来,运行.exe就报错了
**代码报错信息如下:**
D:\modelscope-kws-training-suite-main\dist>test4_luyin.exe
…