带有 S3 位置的 Tensorboard [英] Tensorboard with S3 location

问题描述

我正在尝试使用带有 tensorboard --logdir 命令的 S3 路径并收到以下错误:

I am trying to use S3 path with tensorboard --logdir command and getting the below error:

tensorflow.python.framework.errors_impl.UnimplementedError:文件系统方案s3"未实现(文件:s3://bucket/repo/dummyExperiments")

tensorflow.python.framework.errors_impl.UnimplementedError: File system scheme 's3' not implemented (file: 's3://bucket/repo/dummyExperiments')

即使在环境变量中设置 S3_REGION 后，它也不适合我.
我的张量板版本是 1.7.0.
使用的命令:tensorboard --logdir=s3://repo/dummyExperiments

It is not working for me even after setting S3_REGION in environment variables.
My tensorboard version is 1.7.0.
Command used : tensorboard --logdir=s3://repo/dummyExperiments

我试图从帖子 tensorboard logdir with s3 path 中获取一些输入，但是这不是很有帮助.

I tried to get some input from the post tensorboard logdir with s3 path but it was not very helpful.

tensorboard 是否支持 S3?

Does tensorboard support S3?

推荐答案

是的.Tensorboard 确实支持 S3.

Yes. Tensorboard does support S3.

我创建了一个 EC2 实例并在 ubuntu 机器上安装了 tensorflow 1.7.最初，我在使用 S3 位置运行张量板时收到错误Curl 错误代码 6".但后来我安装了旧的 tensorflow 1.5 版，即使没有设置任何环境变量也能正常工作.

I created an EC2 instance and installed tensorflow 1.7 on the ubuntu machine. Initially I was getting error 'Curl error code 6' on running tensorboard with S3 location. But later I installed old tensorflow version 1.5 and it worked even without setting any environment variables.

不知何故，最新的 1.7 版本与 S3 不兼容并引发错误.

Somehow the latest 1.7 version is not compatible with S3 and throws error.

注意:在 tensorboard --logdir=s3://repo/dummyExperiments 之前，请确保您的服务器已连接到 S3.

NOTE: Before tensorboard --logdir=s3://repo/dummyExperiments , make sure that your server is connected to S3.

这篇关于带有 S3 位置的 Tensorboard的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！