-
Notifications
You must be signed in to change notification settings - Fork 1.8k
TensorFlow 2.0 MNIST example, without IT #1790
Conversation
* Add MNIST example for TensorFlow v2.x * unlock tf version * Print INFO log in "standalone" mode * init on importing * fix intermendiate number * simpler metric reporting and fix ut * try to fix ut * update doc and re-trigger pipeline
cpuNum: 1 | ||
memoryMB: 8196 | ||
#The docker image to run nni job on pai | ||
image: msranni/nni:latest |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Tensorflow 1.x is installed in msranni/nni right?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think so. Because nobody is upgrading it.
I'll remove this config file.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So is it appropriate for us to put an example here that doesn't even work?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I removed docker-based config files. Now this example does work, but only supports limited training services.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Because distributed platforms are not supported, I moved this example down in docs...
Some example trials depend on
tensorflow.examples
, which is not available intf.compat.v1
.So let integration test environment stick to TF 1.x for now.