未验证 提交 432d2b5d 编写于 作者: T Tao Luo 提交者: GitHub

Merge pull request #8105 from qiaohaijun/patch-1

for readability
......@@ -15,10 +15,14 @@ PADDLE_CLUSTER_TRAIN=True
env
# start pserver
stdbuf -oL nohup paddle pserver --port=$PADDLE_INIT_PORT --ports_num=$PADDLE_INIT_PORTS_NUM \
--ports_num_for_sparse=$PADDLE_INIT_PORTS_NUM_FOR_SPARSE --nics=$NICS \
stdbuf -oL nohup paddle pserver \
--port=$PADDLE_INIT_PORT \
--ports_num=$PADDLE_INIT_PORTS_NUM \
--ports_num_for_sparse=$PADDLE_INIT_PORTS_NUM_FOR_SPARSE \
--nics=$NICS \
--comment=paddle_cluster_pserver \
--num_gradient_servers=$PADDLE_INIT_NUM_GRADIENT_SERVERS &> logs/pserver.log &
--num_gradient_servers=$PADDLE_INIT_NUM_GRADIENT_SERVERS \
&> logs/pserver.log &
# start trainer
# NOTE: train.py will use the above environment variables as configuration
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册