-
按照[http://www.paddlepaddle.org/docs/develop/documentation/zh/howto/usage/cluster/k8s_cn.html](url)给的例子进行k8s单机训练：
1.从docker hub拉下基础镜像：`docker pull paddledev/paddle:cpu-demo-latest`
2.运行这个基础镜像`docker …
ghost updated
6 years ago
-
-
paddle_job.py中PADDLE_INIT_TRAINER_COUNT是指每个trainer使用的cpu或者gpu个数？
下面的PADDLE_INIT_TRAINER_COUNT是指trainer的个数，还是指每个trainer使用的cpu或者gpu个数？
paddle.init(
use_gpu=use_gpu,
train…
-
To install dependency we need to use `glide install --strip-vendor` rather than `glide install`
Related issue: https://github.com/Masterminds/glide/issues/709
-
1. Rename `api` into `edl/resource`.
1. Rename function [`api.ConfigureClient`](https://github.com/PaddlePaddle/cloud/blob/dcebcdf820d4a7aa13e815e75d3a17f705794ec1/go/api/register.go#L27) into `edl…
-
请问目前是否支持由master将切分好的数据按存在的trainer进程数量分发？
如果可以，是需要自己部署一套paddle-cloud吗？
谢谢！
-
The job mnist1 has already been deleted, but controller always fetch the information:
```
time="2017-10-26T13:34:37Z" level=info msg="Scaling plan: map[mnist1:0]"
time="2017-10-26T13:34:37Z" level=…
-
rpc client should try to establish connection in period while rpc server is not started. Following is the log:
init fluid.framework.default_startup_program
E0619 09:11:08.576251 159 grpc_client.cc…
-
-