Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Text2SQL-BASELINE 训练速度慢 #308

Open
funs690 opened this issue Apr 19, 2023 · 4 comments
Open

Text2SQL-BASELINE 训练速度慢 #308

funs690 opened this issue Apr 19, 2023 · 4 comments

Comments

@funs690
Copy link

funs690 commented Apr 19, 2023

paddlepaddle-gpu 2.3.2 , GPU A6000, 数据集合3400左右

参数使用默认(batch_size = 16)单轮次训练在2400S,跑完30个轮次需要一天左右的时间,尝试调整batch_size,32/64/128等,实际只有内存占用升高,GPU使用率并无明显提升,训练用时相差不大,该如何调整参数来提高GPU的使用率以提升训练效率?

另外,数据集加载预处理也比较耗时,在资源充足的情况下无法跑满,效率较低,近3w条数据的情况下,光数据集预加载就需要耗时在15min左右!

@fancyerii
Copy link

请问一下你安装ERNIE有没有遇到这个问题:

mkdir third
cd third
git clone https://github.com/PaddlePaddle/ERNIE.git
cd ERNIE && git checkout develop

然后运行下面的脚本出错:

bash data/download_ernie1.0.sh

好像ernie的代码库没有这个脚本了。

@funs690
Copy link
Author

funs690 commented Apr 26, 2023

路径为: ./data/download_ernie1.0.sh
github路径: https://github.com/PaddlePaddle/Research/blob/master/NLP/Text2SQL-BASELINE/data/download_ernie1.0.sh

如果下载不下俩就新建个bash脚本,命令如下:

cd dirname $0

model_files_path="ernie/ernie_1.0_base_ch"

#get pretrained ernie1.0 model params
wget --no-check-certificate https://ernie-github.cdn.bcebos.com/model-ernie1.0.1.tar.gz
if [ ! -d $model_files_path ]; then
mkdir -p $model_files_path
fi
tar xzf model-ernie1.0.1.tar.gz -C $model_files_path
rm model-ernie1.0.1.tar.gz

@fancyerii
Copy link

搞定,谢谢!

@yohohohoho
Copy link

同学,你后来有解决这个问题吗?我这边训练也发现同样的问题。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants